As we run, drive, bike, and fly, we leave behind telltale marks of our movements on Earthif you know where to look. Physical tracks, thermal signatures, and chemical traces can reveal where weve been. But another type of trail we leave comes from the radio signals emitted by the cars, planes, trains, and boats we use. On airplanes, technology called ADS-B (Automatic Dependent SurveillanceBroadcast) provides real-time location, identification, speed, and orientation data. For ships at sea, that function is performed by the AIS (Automatic Identification System). Operating at 161.975 and 162.025 megahertz, AIS transmitters broadcast a ships identification number, name, call sign, length and beam, type, and antenna location every six minutes. Ship location, position time stamp, and direction are transmitted more frequently. The primary purpose of AIS is maritime safetyit helps prevent collisions, assists in rescues, and provides insight into the impact of ship traffic on marine life. US Coast Guard regulations say that generally, private boats under 65 feet in length are not required to use AIS, but most commercial vessels are. Unlike ADS-B in planes, AIS can be turned off only in rare circumstances. A variety of sectors use AIS data for many different applications, including monitoring ship traffic to avoid disruption of undersea internet cables, identifying whale strikes, and studying the footprint of underwater noise. Using the US National Oceanic and Atmospheric Associations Marine Cadastre tool, you can download 16 years of detailed daily ship movements, as well as transit count maps generated from a years worth of data showing each ships accumulated paths. The data is collected entirely from ground-based stations along the US coasts. I downloaded all of 2023s transit count maps and loaded them up in geographic information system software called QGIS to visualize this year of marine traffic. The maps are abstract and electric. With landmasses removed, the ship traces resemble long-exposure photos of sparklers, high-energy particle collisions, or strands of fiber-optic wire. Victoria, British Columbia, and Seattle.DATA: NOAA; MAP: JON KEEGAN / BEAUTIFUL PUBLIC DATA Lake HuronDATA: NOAA; MAP: JON KEEGAN / BEAUTIFUL PUBLIC DATA Savannah, GeorgiaDATA: NOAA; MAP: JON KEEGAN / BEAUTIFUL PUBLIC DATA LouisianaDATA: NOAA; MAP: JON KEEGAN / BEAUTIFUL PUBLIC DATA Zooming in on these maps, you might see strange geometric patterns of perfect circles, or lines in a grid. Some of these are fishing grounds, others are scientific surveys mapping the seafloor, and others represent boats going to and from offshore oil rigs, especially off Louisianas gulf coast. Hiding in plain sight Having a global, near-real-time system for tracking the precise movements of all ships at sea sounds like a great innovationunless youre trying to keep your ships movements and cargoes secret. In 2023, Bloomberg investigated how Russia evaded sanctions on its oil exports after the invasion of Ukraine by spoofingtransmitting fake AIS datato mislead observers. Tracking a fleet of rusting ships of questionable seaworthiness, reporters compared AIS data with what they actually saw on the seaand discovered that the ships werent where the data said they were. Monitoring the fishing industry Clusters of fishing vessels gravitating toward known fishing grounds create some of the most interesting patterns on the maps. Global Fishing Watch is an international nonprofit that uses AIS to monitor the fishing industry, seeking to protect marine life from overfishing. But it says that only 2% of fishing vessels use AIS transmitters. The organization, which is backed by Google, the ocean conservation group Oceana, and the satellite imagery company SkyTruth, combines AIS data with satellite imagery and uses machine learning to classify the types of fishing technology being used. In a press release announcing the creation of Global Fishing Watch, John Amos, the president and founder of SkyTruth, said: So much of what happens out on the high seas is invisible, and that has been a huge barrier to understanding and showing the world whats at stake for the ocean. A version of this story appeared in Beautiful Public Data (beautifulpublicdata.com), a newsletter that curates visually interesting datasets collected by government agencies.

The worst technologies of 2024. The future of mixed reality. AIs impact on the climate. These are just a few of the topics we covered this year in MIT Technology Reviews monthly event series, Roundtables. The series offers a unique opportunity to hear straight from our reporters and editors about what's next for emerging technologies. Available exclusively for subscribers, these 30-minute online discussions provide insights, analysis, and perspectives on timely topics such as gene editing and smart glasses. Roundtables is also a chance for subscribers to ask questions about the latest technologies and learn more about their impact directly from our experts and guests. Subscribers can access recordings of past sessionsabout EVs in China, climate-friendly food, CRISPR babies, and AI hardware. To access the library, simply log in with your subscription or subscribe now to save 25% and unlock access to the entire series. Here are some highlights from this year in Roundtables: The Worst Technology Failures of 2024 MIT Technology Review publishes an annual list of the worst technologies of the yearchronicling flops, failures, and other mishaps. The 2024 list was unveiled in December by executive editor Niall Firth and senior editor for biomedicine Antonio Regalado. They had a lively discussion about each of the eight items on this listand what we can learn from these fiascos. Whats Next for Mixed Reality: Glasses, Goggles, and More This year brought many new developments in one particular consumer device category: smart glasses. After years of development, new augmented-reality specs from several companies made their debut. Editor in chief Mat Honan and AI hardware reporter James ODonnell talked about where its all heading. Putting AIs Climate Impact into Perspective The rise of AI comes with a growing carbon footprint and greater demand for electricity. Analysts project that AI could drive up data centers energy consumption by 160% this decade. So how worried should we be? Editor at large David Rotman, senior AI reporter Melissa Heikkil, and senior editor for energy James Temple explored the energy trade-offs involved in AI. CRISPR Babies: Six years later Gene editing can correct or improve the DNA of human embryos, potentially opening the door to the technological evolution of our species. But in 2018, a premature attempt to use the technology this way led to a prison term for He Jiankui, the researcher involved. Editor in chief Mat Honan and senior editor for biomedicine Antonio Regalado had a conversation with He, a biophysicist and the creator of the first gene-edited humans, to revisit this controversial technology and the future of editing in IVF clinics. Why Thermal Batteries Are So Hot Right Now Thermal batteries could be a key part of cleaning up heavy industry. Executive editor Amy Nordrum and senior climate reporter Casey Crownhart told us what we can expect next from this emerging technologywhich was also voted the 11th breakthrough technology of 2024 by our readers.

Another year is coming to a close, so lets look back at the MIT Technology Review stories that resonated most with you, our readers. We published hundreds of stories in 2024, about AI, climate tech, biotech, robotics, space, and more. There were six new issues of our magazine, on themes including food, play, and hidden worlds. We launched two newsletters, to share tech industry analysis from our editor in chief and to step people through the basics of AI. And we hosted 11 exclusive conversations with our editors and experts in our subscriber-only event series, Roundtables. What did people enjoy most? Heres a quick look at some of the stories that performed best with our audience: 10 Breakthrough Technologies of 2024 Every year as we compile this annual list, we look for promising technologies poised to have a real impact on the world. It represents the advances that we think matter most, and the 2024 edition included weight-loss drugs, chiplets, and the first gene-editing treatment. The 2025 list is dropping in early January. To find out what made the cut, join us for a special live Roundtables event, Unveiling the 10 Breakthrough Technologies of 2025, on Friday, January 3, at 12:30 p.m. ET. This is a subscriber-only event. Register to attend or subscribe for access.) What is AI? Everyone thinks they know, but no one can agree. Senior editor Will Douglas Heaven explored the problem in this in-depth feature storyand explained why it matters for all of our futures. He covers the origins of modern AI and digs into the ongoing debate among experts about this technologys capabilities and potential. The AI Hype Index Theres no denying AI moves fast, and it can be hard to know whats worth your attention. Thats why we started plotting everything you need to know about the state of AI in a new matrix, along axes that run from Hype to Real and Doom to Utopia. What are AI agents? Major tech companies are now developing AI tools that can do more complex tasks, like sending emails or booking plane tickets, on your behalf. Heres how they will work. Super-efficient solar cells: 10 Breakthrough Technologies 2024 Solar cells that combine traditional silicon with cutting-edge perovskites could push the efficiency of solar panels to new heights. Thats why we put them on our list of the 10 Breakthrough Technologies of 2024. Happy birthday, baby! What the future holds for those born today As part of our 125th anniversary issue, contributor Kara Platoni spoke with a dozen experts to sketch out how technology might influence the life of someone born today over the next 125 years. The messy quest to replace drugs with electricity In the 2010s, the field of electroceuticals was born, attracting much fanfare and investment. Contributor Sally Adee explored how the field fizzled and how its being revived as an effort to turn gene expression on and off with electric fields. 15 Climate Tech Companies to Watch For the second annual edition of this list, our reporters and editors chose 15 companies from around the world that we think have the best shot at making a difference on climate change. Weight-loss drugs: 10 Breakthrough Technologies 2024 Drugs like Wegovy and Mounjaro have quickly become embedded into American life. In 2024, they even earned a place on our 10 Breakthrough Technologies list. The long-term implications are unknown, but plenty of people are using semaglutides anyway, and many lose around 15% of their body weight. Don't miss out on even more emerging technology coverage and subscriber-only stories.Subscribe today for unlimited accessto expert insights that you can't find anywhere else.

This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here. Heres a question. Imagine that, for $15,000, you could purchase a robot to pitch in with all the mundane tasks in your household. The catch (aside from the price tag) is that for 80% of those tasks, the robots AI training isnt good enough for it to act on its own. Instead, its aided by a remote assistant working from the Philippines to help it navigate your home and clear your table or put away groceries. Would you want one? Thats the question at the center of my story for our magazine, published online today, on whether we will trust humanoid robots enough to welcome them into our most private spaces, particularly if theyre part of an asymmetric labor arrangement in which workers in low-wage countries perform physical tasks for us in our homes through robot interfaces. In the piece, I wrote about one robotics company called Prosper and its massive effortbringing in former Pixar designers and professional butlersto design a trustworthy household robot named Alfie. Its quite a ride. Read the story here. Theres one larger question that the story raises, though, about just how profound a shift in labor dynamics robotics could bring in the coming years. For decades, robots have found success on assembly lines and in other somewhat predictable environments. Then, in the last couple of years, robots started being able to learn tasks more quickly thanks to AI, and that has broadened their applications to tasks in more chaotic settings, like picking orders in warehouses. But a growing number of well-funded companies are pushing for an even more monumental shift. Prosper and others are betting that they dont have to build a perfect robot that can do everything on its own. Instead, they can build one thats pretty good, but receives help from remote operators anywhere in the world. If that works well enough, theyre hoping to bring robots into jobs that most of us would have guessed couldn't be automated: the work of hotel housekeepers, care providers in hospitals, or domestic help. Almost any indoor physical labor is on the table, Prospers founder and CEO, Shariq Hashme, told me. Until now, weve mostly thought about automation and outsourcing as two separate forces that can affect the labor market. Jobs might be outsourced overseas or lost to automation, but not both. A job that couldnt be sent offshore and could not yet be fully automated by machines, like cleaning a hotel room, wasnt going anywhere. Now, advancements in robotics are promising that employers can outsource such a job to low-wage countries without needing the technology to fully automate it. Its a tall order, to be clear. Robots, as advanced as theyve gotten, may find it difficult to move around complex environments like hotels and hospitals, even with assistance. That will take years to change. However, robots will only get more nimble, as will the systems that enable them to be controlled from halfway around the world. Eventually, the bets made by these companies may pay off. What would that mean? One, the labor movements battle with AIwhich this year has focused its attention on automation at ports and generative AIs theft of artists workwill have a whole new battle to fight. It wont just be dock workers, delivery drivers, and actors seeking contracts to protect their jobs from automationit will be hospitality and domestic workers too, along with many others. Second, our expectations of privacy would radically shift. People buying those hypothetical household robots would have to be comfortable with the idea that someone that they have never met is seeing their dirty laundryliterally and figuratively. Some of those changes might happen sooner rather than later. For robots to learn how to navigate places effectively, they need training data, and this year has already seen a race to collect new data sets to help them learn. To achieve their ambitions for teleoperated robots, companies will expand their search for training data to hospitals, workplaces, hotels, and more. Now read the rest of The Algorithm Deeper Learning This is where the data to build AI comes from AI developers often dont really know or share much about the sources of the data they are using, and the Data Provenance Initiative, a group of over 50 researchers from both academia and industry, wanted to fix that. They dug into 4,000 public data sets spanning over 600 languages, 67 countries, and three decades to understand whats feeding todays top AI models, and how that will affect the rest of us. Why it matters: AI is being incorporated into everything, and what goes into the AI models determines what comes out. However, the team found that AI's data practices risk concentrating power overwhelmingly in the hands of a few dominant technology companies, a shift from how AI models were being trained just a decade ago. Over 90% of the data sets that the researchers analyzed came from Europe and North America, and over 70% of data for both speech and image data sets comes from YouTube. This concentration means that AI models are unlikely to capture all the nuances of humanity and all the ways that we exist, says Sara Hooker, a researcher involved in the project. Read more from MelissaHeikkil. Bits and Bytes In the shadows of Arizonas data center boom, thousands live without power As new research shows that AIs emissions have soared, Arizona is expanding plans for AI data centers while rejecting plans to finally provide electricity to parts of the Navajo Nations land. (Washington Post) AI is changing how we study bird migration After decades of frustration, machine-learning tools are unlocking a treasure trove of acoustic data for ecologists. (MIT Technology Review) OpenAI unveils a more advanced reasoning model in race with Google The new o3 model, unveiled during a livestreamed event on Friday, spends more time computing an answer before responding to user queries, with the goal of solving more complex multi-step problems. (Bloomberg) How your car might be making roads safer Researchers say data from long-haul trucks and General Motors cars is critical for addressing traffic congestion and road safety. Data privacy experts have concerns. (New York Times)

Dan Huttenlocher, SM 84, PhD 88, leads the way up to the eighth floor of Building 45, the recently completed headquarters of the MIT Schwarzman College of Computing. Theres an amazing view of the Great Dome here, he says, pointing out a panoramic view of campus and the Boston skyline beyond. The floor features a high-end event space with an outdoor terrace and room for nearly 350 people. But it also serves an additional purposeluring people into the building, which opened last January. The event space wasnt in the original building plan, says Huttenlocher, Schwarzmans inaugural dean, but the point of the building is to be a nexus, bringing people across campus together. Launched in 201920, Schwarzman is MITs only college, so called because it cuts across the Institutes five schools in a new effort to integrate advanced computing and artificial intelligence into all areas of study. We want to do two things: ensure that MIT stays at the forefront of computer science, AI research, and education, Huttenlocher says, and infuse the forefront of computing into disciplines across MIT. He adds that safety and ethical considerations are also critical. To that end, the college now encompasses multiple existing labs and centers, including the Computer Science and Artificial Intelligence Laboratory (CSAIL), and multiple academic units, including the Department of Electrical Engineering and Computer Science. (EECSwhich was reorganized into the overlapping subunits of electrical engineering, computer science, and artificial intelligence and decision-makingis now part of both the college and the School of Engineering.) At the same time, the college has embarked on a plan to hire 50 new faculty members, half of whom will have shared appointments in other departments across all five schools to create a true Institute-wide entity. Those faculty memberstwo-thirds of whom have already been hiredwill conduct research at the boundaries of advanced computing and AI. We want to do two things: ensure that MIT stays at the forefront of computer science, AI research, and education and infuse the forefront of computing into disciplines across MIT. Dan Huttenlocher The new faculty members have already begun helping the college respond to an undeniable reality facing many students: Theyve been overwhelmingly drawn to advanced computing tools, yet computer science classes are often too technical for nonmajors who want to apply those tools in other disciplines. And for students in other majors, it can be tricky to fit computer science classes into their schedules. Meanwhile, the appetite for computer science education is so great that nearly half of MITs undergraduates major in EECS, voting with their feet about the importance of computing. Graduate-level classes on deep learning and machine vision are among the largest on campus, with over 500 students each. And a blended major in cognition and computing has almost four times as many enrollees as brain and cognitive sciences. Weve been calling these students computing bilinguals, Huttenlocher says, and the college aims to make sure that MIT students, whatever their field, are fluent in the language of computing. As we change the landscape, he says, its not about seeing computing as a tool in service of a particular discipline, or a discipline in the service of computing, but asking: How can we bring these things together to forge something new? The college has been the hub of this experiment, sponsoring over a dozen new courses that integrate computing with other disciplines, and it provides a variety of spaces that bring people together for conversations about the future of computing at MIT. More than just a nexus for computing on campus, the college has also positioned itself as a broad-based leader on AI, presenting policy briefs to Congress and the White House about how to manage the pressing ethical and political concerns raised by the rapidly evolving technology. Right now, digital technologies are changing every aspect of our lives with breakneck speed, says Asu Ozdaglar, SM 98, PhD 03, EECS department head and Schwarzmans deputy dean of academics. The college is MITs response to the ongoing digital transformation of our society. Huttenlocher, who also holds the title of Henry Ellis Warren (1894) Professor of Electrical Engineering and Computer Science and coauthored the book The Age of AI: And Our Human Future with Henry Kissinger and Eric Schmidt, has long been exploring such issues. He started programming computers back in middle school in Connecticut in the 1970s on an ASR 33 teletype machine, and eventually he studied at the University of Michigan as a double major in cognitive psychology and computer science, exploring speech recognition and visual perception. AI work back then was relatively disconnected from the physical world, he says. Being interested in the perceptual side of things was kind of an outlier for what was going on in AI then. When he looked at grad schools in the 1980s, only MIT, Carnegie Mellon, and Stanford were doing significant work in AI, he says: I applied to those three schools and figured if it didnt work out, Id get a job. It worked out, of course. He headed to Cambridge and gravitated to MITs AI Lab in Technology Square, where he first worked on speech recognition and then transitioned into computer vision, at the time still in its infancy. After earning his PhD, he served simultaneously as a computer science professor at Cornell and a researcher at Xerox PARC, flying between New York and the burgeoning Silicon Valley, where he worked on computer vision for the digital transformation of copiers and scanners. In academia, you have more curiosity-driven research projects, where in the corporate world you have the opportunity to build things people will actually use, he says. Ive spent my career moving back and forth between them. Along the way, Huttenlocher gained administrative experience as well. He was a longtime board member and eventual chair of the MacArthur Foundation, and he also helped launch Cornell Tech, the universitys New York Citybased graduate school for business, law, and technology, serving as its first dean and vice provost. When Stephen Schwarzman, CEO of the investment firm Blackstone Group, gave $350 million to MIT to establish a college of computing in 2018, he was eager to return to the Institute to lead it. The fact that MIT was making a bold commitment to become a broad-based leader in the AI-driven ageand that it was cutting across all of its schoolswas exciting, he says. Schwarzman College took shape through task forces involving more than 100 MIT faculty members. By the fall of 2019 a plan had been nailed down, and Huttenlocher was in place as director with EECS head Ozdaglar named deputy dean of academics. I never believed that everybody wants to do computer science at MIT, she says. Students come in with a lot of passions, and its our responsibility to educate these bilinguals, so they are fluent in their own discipline but also able to use these advanced frontiers of computing. Ozdaglars background is in using machine learning to optimize communications, transportation, and control systems. Recently she has become interested in applying machine-learning algorithms to social media, examining how the choices people make when sharing content affect the informationand misinformationrecommended to them. This work builds on her longstanding interdisciplinary collaborations in the social sciences, including collaborations with her husband, economics professor (and recent Nobel laureate) Daron Acemoglu. I strongly feel that to really address the important questions in society, these old department or disciplinary silos arent adequate anymore, she says. The college has enabled me to work much more broadly across MIT and share all that Ive learned. Ozdaglar has been a driving force behind faculty hiring for the college, working with 18 departments to bring on dozens of scholars at the forefront of computing. In some ways, she says, its been a challenge to integrate the new hires into existing disciplines. We have to keep teaching what weve been teaching for tens or hundreds of years, so change is hard and slow, she says. But she has also noticed a palpable excitement about the new tools. Already, the college has brought in more than 30 new faculty members in four broad areas: climate and computing; human and natural intelligence; humanistic and social sciences; and AI for scientific discovery. In each case, they receive an academic home in another department, as well as an appointment, and often lab space, within the college. Asu Ozdaglar, SM 98, PhD 03, Schwarzmans deputy dean of academics, in the lobby of the new headquarters building. That commitment to interdisciplinary work has been built into every aspect of the new headquarters. Most buildings at MIT come across as feeling pretty monolithic, Huttenlocher says as he leads the way along brightly lit hallways and common spaces with large walls of glass looking out onto Vassar Street. We wanted to make this feel as open and accessible as possible. While the Institutes high-end computing takes place mostly at a massive computing center in Holyoke, about 90 miles away in Western Massachusetts, the building is honeycombed with labs and communal workspaces, all made light and airy with glass and natural blond wood. Along the halls, open doorways offer enticing glimpses of such things as a giant robot hanging from a ceiling amid a tangle of wires. Lab and office space for faculty research groups working on related problemswho might be from, say, CSAIL and LIDSis interspersed on the same floor to encourage interaction and collaboration. Its great because it builds connections across labs, Huttenlocher says. Even the conference room does not belong to either the lab or the college, so people actually have to collaborate to use it. Another dedicated space is available six months at a time, by application, for special collaborative projects. The first group to use it, last spring, focused on bringing computation to the climate challenge. To make sure undergrads use the building too, theres a classroom and a 250-seat lecture hall, which now hosts classic Course 6 classes (such as Intro to Machine Learning) as well as new multidiscipline classes. A soaring central lobby lined with comfortable booths and modular furniture is ready-made for study sessions. For some of the new faculty, working at the college is a welcome change from previous academic experiences in which they often felt caught between disciplines. The intersection of climate sustainability and AI was nascent when I started my PhD in 2015, says Sherrie Wang, an assistant professor with a shared appointment in mechanical engineering and the Institute for Data, Systems, and Society, who is principal investigator of the Earth Intelligence Lab. When she hit the job market in 2022, it still wasnt clear which department shed be in. Now a part of Schwarzmans climate cluster, she says her work uses machine learning to analyze satellite data, examining crop distribution and agricultural practices across the world. Its great to have a cohort of people who have similar philosophical motivations in applying these tools to real-world problems, she says. At the same time, were pushing the tools forward as well. AI impact papersIn the fall of 2023, MIT began providing seed funding to teams of MIT faculty and researchers to explore how generative AI will transform peoples lives and work.As generative AI evolves at an exceptionally rapid pace, MIT has a responsibility to help humanity pursue a future of AI innovation that is broadly beneficial and mitigates potential harm, President Sally Kornbluth observed when announcing the publication of the first set of resulting white papers. A deep understanding of the societal impact of AI is a vital part of this effort, and MIT faculty have an extraordinary breadth of knowledge and insight to contribute.To date, MIT has published preprints of 39 white papers on a wide range of topics. Browse them all at https://mit-genai.pubpub.org/. Among other researchers, she plans to collaborate with Sara Beery, a CSAIL professor who analyzes vast troves of visual, auditory, and other data from a diverse range of sensors around the world to better understand how climate change is affecting distribution of species. AI can be successful in helping human experts efficiently process terabytes and petabytes of data so they can make informed management decisions in real time rather than five years later, says Beery, who was drawn to the colleges unique hybrid nature. We need a new generation of researchers that frame their work by bringing different types of knowledge together. At Schwarzman, there is a clear vision that this type of work is going to be necessary to solve these big, essential problems. Beery is now working to develop a class in machine learning and sustainability with two other new faculty members in the climate cluster: Abigail Bodner, an assistant professor in EECS and Earth, Atmospheric, and Planetary Sciences (whose work uses AI to analyze fluid dynamics), and Priya Donti, assistant professor in EECS and LIDS (who uses AI and computing to optimize integration of renewable energy into power grids). Theres already a core course on AI and machine learningan on-ramp for people without prior exposure who want to gain those fundamentals, says Donti. The new class would be for those who want to study advanced AI/ML topics within the context of sustainability-related disciplines, including power systems, biodiversity, and climate science. The class on machine learning and sustainability would be part of Common Ground for Computer Education, an initiative cochaired by Ozdaglar and involving several dozen faculty members across MIT to develop new classes integrating advanced computing with other disciplines. So far, says Ozdaglar, it has generated more than a dozen new courses. One machine-learning class developed with input from nine departments provides exposure to a variety of practical applications for AI algorithms. Another collaboration, between computer science and urban studies, uses data visualization to address housing issues and other societal challenges. Julia Schneider 26, a double major in AI and mathematics, took the Common Ground class on optimization methods, which she says demonstrated how computer science concepts like shortest-path algorithms and reinforcement learning could be applied in other areas, such as economics and business analytics. She adds that she values such classes because they blend her two areas of study and highlight multidisciplinary opportunities. Even faculty who are leading researchers in this area say I cant read fast enough to keep up with whats going on. Dan Huttenlocher Natasha Hirt 23, MEng 23, came to MIT thinking that computer science was peripheral to her major in architecture and urban planning. Then she took a course with building technology professor Caitlin Mueller on structural optimization and designand it changed the trajectory of her MIT career. That led her to Interactive Data Visualization and Society, a Common Ground class, and several interdisciplinary classes combining computer science and field-specific knowledge. She says these provided the perfect introduction to algorithms without delving too much into math or coding,giving her enough working knowledge to set up models correctly and understand how things can go wrong. They are teaching you what an engine is, what it looks like, and how it works without actually requiring you to know how to build an engine from scratch, she says, though she adds that the classes also gave her the opportunity to tinker with the engine. Shes now working on masters degrees in both building technology and computation science and engineering, focusing on making buildings more sustainable by using computational tools to design novel, less material-intensive structures. She says that Common Ground facilitates an environment where students dont have to be computer science majors to learn the computational skills they need to succeed in their fields. And thats the intent. My hope is that this new way of thinking and these educational innovations will have an impact both nationally and globally, Ozdaglar says. The same goes for recent papers MIT has commissioned, both on AI and public policy and on applications of generative AI. As generative AI has spread through many realms of society, it has become an ethical minefield, giving rise to problems from intellectual-property theft to deepfakes. The likely consequence has been to both over- and under-regulate AI, because the understanding isnt there, Huttenlocher says. But the technology has developed so rapidly its been nearly impossible for policymakers to keep up. Even faculty who are leading researchers in this area say I cant read fast enough to keep up with whats going on, Huttenlocher says, so that heightens the challengeand the need. The college has responded by engaging faculty at the cutting edge of their disciplines to issue policy briefs for government leaders. First was a general framework written in the fall of 2023 by Huttenlocher, Ozdaglar, and the head of MITs DC office, David Goldston, with input from more than a dozen MIT faculty members. The brief spells out essential tasks for helping the US maintain its AI leadership, as well as crucial considerations for regulation. The college followed that up with a policy brief by EECS faculty specifically focusing on large language models such as ChatGPT. Others dealt with AIs impact on the workforce, the effectiveness of labeling AI content, and AI in education. Along with the written documents, faculty have briefed congressional committees and federal agencies in person to get the information directly into the hands of policymakers. The question has been How do we take MITs specific academic knowledge and put it into a form thats accessible? Huttenlocher says. On a parallel track, in July of 2023 President Sally Kornbluth and Provost Cynthia Barnhart, SM 86, PhD 88, issued a call for papers by MIT faculty and researchers to articulate effective road maps, policy recommendations, and calls for action across the broad domain of generative AI. Huttenlocher and Ozdaglar played a key role in evaluating the 75 proposals that came in. Ultimately, 27 proposalsexploring the implications of generative AI for such areas as financial advice, music discovery, and sustainabilitywere selected from interdisciplinary teams of authors representing all five schools. Each of the 27 teams received between $50,000 and $70,000 in seed funds to research and write 10-page impact papers, which were due by December 2023. Given the enthusiastic response, MIT sent out another call in the fall of 2023, resulting in an additional 53 proposals, with 16 selected in March, on topics including visual art, drug discovery, and privacy. As with the policy briefs, Huttenlocher says, we are trying to provide the fresher information an active researcher in the field would have, presented in a way that a broader audience can understand. Even in the short time the college has been active, Huttenlocher and Ozdaglar have begun to see its effects. Were seeing departments starting to change some of the ways they are hiring around degree programs because of interactions with the college, Huttenlocher says. There is such a huge acceleration of AI in the worldits getting them to think with some urgency in doing this. Whether through faculty hiring, new courses, policy papers, or just the existence of a space for high-level discussions about computing that had no natural home before, Huttenlocher says, the college hopes to invite the MIT community into a deeper discussion of how AI and other advanced computing tools can augment academic activities around campus.MIT has long been a leader in the development of AI, and for many years it has continued to innovate at the cutting edge of the field. With the colleges leadership, the Institute is in a position to continue innovating and to guide the future of the technology more broadly. The next step, says Ozdaglar, is to take that impact out into the world.

THE CULT OF THE FOUNDER. THE CULT OF THE TECH GENIUS. Beware: Silicon Valleys cultists want to turn you into a disruptive deviant. Techs cult of the founder bounces back. Silicon Valleys Strange, Apocalyptic Cults. How the cult of personality and tech-bro culture is killing technology. Company or cult? Is your corporate culture cultish? The Cult of Company Culture Is Back. But Do Tech Workers Even Want Perks Anymore? 10 tech gadgets with a cult following on Amazonand why theyre worth it. 13 steps to developing a cult-like company culture. The headlines seem to write themselves (if that clich is allowed anymore in the age of ChatGPT and generative AI). Tech is culty. But that is a metaphor, right? Right?! When I first saw Michael Saylors Twitter account, I wasnt sure. Saylor is an entrepreneur, tech executive, and former billionaire. Once reportedly the richest man in the Washington, DC, area, he lost most of his $7 billion net worth in 2000 when, in his mid-30s, he reached a settlement with the US Securities and Exchange Commission after it brought charges against him and two of his colleagues at a company called MicroStrategy for inaccurate reporting of their financial results. But I had no idea who he was back then. In 2021 Saylor started showing up in my Twitter feed. His profile picture showed a man with chiseled features, silver hair, and stubble sitting in a power pose and looking directly into the camera, a black dress shirt unbuttoned to display a generous amount of his neck. It was a typical tech entrepreneurs publicity shot except for the lightning bolts blasting from his eyes, and the golden halo crown. Then there were his tweets: #Bitcoin is Truth. #Bitcoin is For All Mankind. #Bitcoin is Different. Trust the Timechain. Fiat [government-backed currency] is immoral. #Bitcoin is immortal. #Bitcoin is a shining city in cyberspace, waiting for you. #Bitcoin is the heartbeat of Planet Earth. As MITs humanist chaplain, I follow a lot of ministers, rabbis, imams, and monks online. Very few religious leaders would dare to be this religious on social media. They know that few of their readers want to see such hubris. Why, then, does there seem to be an audience for this seemingly cultish behavior from a cryptocurrency salesman? Are tech leaders like Saylor leading actual cults? According to Bretton Putter, an expert on startups and CEO of the consulting firm CultureGene, this neednt be a major concern: Its pretty much impossible, Putter writes, for a business to become a full-blown cult. And if a tech company or other business happens to resemble a cult, that might just be a good thing, he argues: If you succeed in building a cultlike culture similar to the way that Apple, Tesla, Zappos, Southwest Airlines, Nordstrom, and Harley-Davidson have, you will experience loyalty, dedication, and commitment from your employees (and customers) that is way beyond the norm. Are the cultlike aspects of tech companies really that benign? Or should we be worried? To find the answer, I interviewed Steve Hassan, a top expert on exit counseling, or helping people escape destructive cults. At age 19, while he was studying poetry at Queens College in New York City in the early 1970s, Hassan was recruited into the Unification Churchthe famously manipulative cult also known as the Moonies. Over his next 27 months as a member of the church, Hassan helped with its fundraising, recruiting, and political efforts, which involved personally meeting with the cult leader Sun Myung Moon multiple times. He lived in communal housing, slept only a few hours a night, and sold carnations on street corners seven days a week for no pay. He was told to drop out of college and turn his bank account over to the church. In 1976, he fell asleep at the wheel while driving a Moonie fundraising van and drove into the back of a tractor-trailer at high speed. He called his sister from the hospital, and his parents hired former members to help deprogram him and extract him from the cult. After the Jonestown mass suicide and murders of 1978 brought attention to the lethal dangers of cult mind control, Hassan founded a nonprofit organization, Ex-Moon Inc. Since then, hes earned a handful of graduate degrees (including a doctorate in the study of cults), started numerous related projects, and written a popular book on how practices with which he is all too familiar have crept into the mainstream of US politics in recent years. (That 2019 book, The Cult of Trump: A Leading Cult Expert Explains How the President Uses Mind Control, seemed even more relevant in early 2024, when a video called God Made Trump went viral across the campaign trail.) Hassan even found himself advising Maryland congressman Jamie Raskin, leader of the second impeachment trial against Donald Trump, in 2021, on how to think and communicate about the cultish aspects of the violent mob of Trump followers who stormed the Capitol on January 6 of that year. I wanted to ask Hassan what he makes of the discourse around tech cults, but first its important to understand how he thinks about cults in the first place. Hassans dissertation was titled The BITE Model of Authoritarian Control: Undue Influence, Thought Reform, Brainwashing, Mind Control, Trafficking, and the Law. The idea was to create a model that could measure cult exploitation and manipulation, or what Hassan and other experts in related fields call undue influence. His BITE model looks to evaluate the ways social groups and institutions attempt to control followers behavior, information access, thoughts, and emotions. Because there is no one quintessential, Platonic definition of a cult, what matters is where a given instance of potential cultishness falls on an influence continuum. In this continuum model, Hassan evaluates the ways in which institutional cultures attempt to influence people. To what extent are individuals allowed to be their authentic selves or required to adopt a false cult identity? Are leaders accountable to others, or do they claim absolute authority? Do organizations encourage growth in the people who participate in them, or do they seek to preserve their own power over all else? While any kind of person or group can struggle with some of the dimensions on Hassans continuum chart (which lists constructive behaviors at one end and destructive behaviors at the other), healthier organizations will tend toward constructive responses more of the time, whereas unhealthier institutionsthose more truly worthy of the cult label in the most negative sensewill tend toward destructive responses such as grandiosity, hate, demands for obedience, elitism, authoritarianism, deceptiveness, or hunger for power. It turns out that there are some real, meaningful similarities between cults and tech, according to Hassan. This is the perfect mind-control device, he told me, holding up his iPhone. He explained that when he joined the Moonies in 1974, cult recruiters had to get information from the victim. Now, he said, users of everyday technologies are sitting ducks: There are 5,000 data points on every voting American in the dark web, and there are companies that will collect and sell that data. The first time Hassan was told about cryptocurrency, he added, it smacked of multilevel marketing to him. The proposition that you can make a fortune in a very short amount of time, with almost no labor, was something he had seen many times in his work. As was the idea that if you become an early investor in such a scheme, youll make more money if you recruit more people to join you. The people who started it are always going to make 99% of the money, Hassan said. And as in the cults that recruited him and continue to recruit the kinds of people who ultimately become his clients, everyone else is going to get burned. All of this would certainly seem to explain why I so frequently hear from people, eager for me to know they are fellow atheists, who tell me to buy some bitcoin because it will rewire my neurons and cure me of the woke mind virus. Of course, it should be noted that some scholars have complained about Hassans work, arguing that brainwashing and mind control are concepts for which there is not sufficient evidence. But Im not claiming that tech uses literal brainwashing, nor is it like when a character in a Scooby-Doo episode hears You are getting very sleepy and then their eyes become squiggles. Hassan probably wouldnt say so either. Companies dont need to go to such extremes to exert undue influence on us, though. And as is clear from the headlines I cited above, a lot of companies have been accused of, or associated with, a bit of cultishness. I wont attempt to evaluate anyones cultish tendencies on a scale of 1 to 10. But I see crypto sales techniques as a particularly good example of cultlike behavior, because if theres one thing cults need to be good at to sustain their existence, its separating people from their wallets. Cryptocurrency has specialized in that to extraordinary effect. Its all a continuum, and it would be hard to find a person whose life is completely devoid of anything cultish, technological or otherwise. But as a culture, we are careening dangerously toward the wrong end of Hassans chart. Or to quote a Michael Saylor tweet, We all stumble in the dark until we see the cyber light. #Bitcoin. Adapted from Tech Agnostic: How Technology Became the Worlds Most Powerful Religion, and Why It Desperately Needs a Reformation. Copyright 2024 by Greg Epstein, the humanist chaplain at MIT. Used with permission of the publisher, MIT Press.

We need a strategy to deal with a hydra. Its Sunday, January 14, 2024, more than 50 hours since the annual MIT Mystery Hunt kicked off at noon on Friday, and Setec Astronomy is one of more than 200 teams racing to solve hundreds of puzzles over three days. The 60-some members of Setec, many of whom are joining remotely from as far away as Australia, are making good progress, even though many of us are running on limited sleep and questionable nutritional decisions. Several of the chalkboards in the Building 2 classroom weve been assigned for our team headquarters are covered in lists of puzzle solutions or messy diagrams charting out theories about how to crack the various challengesall of them constructed, as Mystery Hunt tradition dictates, by the most recent winner, in this case The Team Formerly Known as the Team to Be Named Later. The hydra were dealing with is a metapuzzle: We have to find a way to use the solutions from other puzzles that weve already solved to extract one more answer. If we solve this one, well be rewarded with more puzzles. We know we need to diagram the answers for this round of puzzles as a binary tree. In keeping with the hydra metapuzzles mythological analogue, every time we solve one puzzle, two more branch off until we have a diagram five levels deep. Were still missing answers from several unsolved puzzles that would help us figure out how the diagram works and how to extract an answer to the metapuzzle. The diagram weve drawn, in green chalk, gets more chaotic with every addition, erasure, and annotation we squeeze onto the overcrowded chalkboard. But we can sense that were just one aha! away from a solution. MITs Mystery Hunt has been challenging puzzle enthusiasts every year since Brad Schaefer 78, PhD 83, wrote 12 subclues on a single sheet of paper as a challenge for friends during Independent Activities Period (IAP) in 1981. The answers led solvers to an Indian Head penny he had hidden on campus. Todays Hunts are still built around that basic concept, but what constitutes a challenge has changed over four decades. One of the clues from the original 1981 Hunt is just a missing word in a quote: He that plays the king shall be _____; his majesty shall have tribute of me. Its easy to solve today with Google, but in 1981, even if you knew it was Shakespeare, if you didnt notice the subtle hint that you should look for a character referring to a play within the play, it might have taken a few hours of skimming the Bards collected works to find the answer. The Setec Astronomy team tries to map out whether the human knot theyve gotten themselves into can be untangled.JADE CHONGSATHAPORNPONG 24/MIT TECHNIQUE We add a few more solutions to the hydra diagram over the next few hours. Eventually someone notices that all the answers in the fifth level of the diagram seem to have an odd prevalence of Ls and Rs. This is the aha! moment: They tell us how to navigate the binary tree. From the first node at the top of the tree, we follow the Ls and Rs in the order they appear in each of the 16 solutions on the fifth level. Take the left branch, then right, then left again, landing on a word that starts with H. The second fifth-level answer leads us to a word that starts with E. Repeating the process with all 16 answers spells out an apt way to deal with a hydra: HEADTOHEADBATTLE. (Puzzle solutions are traditionally written in all caps with no spaces or punctuation.) Those of us whove been tackling the puzzle take a moment to enjoy our victory before splitting up to find new puzzles to work on. Some elements of the Mystery Hunt are hard to describe, the kind of must-be-seen ingenuity that also inspires hacks on the Great Dome and any number of above-and-beyond engineering projects showcased around campus every year. Most of the puzzles are utterly unique, although they do often incorporate logic and word problems as well as more mainstream elements like crosswords, sudoku, and Wordle. But almost anything can be turned into a puzzle. For example, chess puzzles might be combined with the card game Magic: The Gathering. Or solvers could be asked to organize a Git repository with 10,000 out-of-order commits (that is, find the correct sequence of 10,000 changes to a file as it was tracked in a version control system), identify duets from musicals, or draw on their knowledge of pop culture trivia. For most of its history, the Mystery Hunt had little official status on campus. By tradition as much as any organizational effort, teams simply showed up in Lobby 7 on the Friday before the Martin Luther King Jr. holiday for the kickoff. In 2014, the MIT Puzzle Club was formed to help provide year-to-year continuity and other support, such as securing rooms for teams to work in and reserving Kresge Auditorium for the opening ceremonies. Puzzle Club also hosts other events, such as mini puzzle hunts and sudoku and logic puzzle competitionswhich Becca Chang 26, the clubs current president, says has helped a lot with outreach to new students or anyone who might be interested in [puzzles]. Technology has enabled the Mystery Hunt to grow and evolve in significant ways, and not just in terms of the kinds of puzzles that are possible. Through the mid-1990s, a single person could take on the responsibility of writing and running the event. Today its a yearlong commitment for the winning team to design the next years Hunt. Doing so requires managing creative output and technological infrastructure that rival those of a small business. Duties include spending thousands of hours writing and testing puzzles, constructing physical puzzles and props, and building a dynamic website that can withstand the huge influx of puzzle-hungry visitors. Todays Hunts are built around a story. Here John Bromels as the god Neptune checks in on Galactic Trendsetters progress to restore the god Pluto after his planet was demoted.JADE CHONGSATHAPORNPONG 24/MIT TECHNIQUE Just organizing a team of solvers can be a major undertaking, especially now that more and more participants are joining remotely. Anjali Tripathi 09, who started the team Im Not a Planet Either in 2015, got her introduction to puzzle hunts through a miniature Mystery Hunt that Simmons Hall runs for first-years. After tackling the main event with the Simmons team on campus as an undergrad, she participated remotely for the first time in 2010. I was abroad in England and still wanted to do Hunt, and I remember how hard that was, she says. The team had no infrastructure for it. Its about connecting with other humans thats why we do it. Erin Rhode 04, whose team name one year was the entire text of Ayn Rands Atlas Shrugged Today, solvers can work together across the room or across a continent. Platforms like Slack and Discord have become indispensable to many teams, which use them for updates and announcements as well as creating separate channels where people can tackle a given puzzle together. Many teams use applications that organize the convoluted deluge of puzzles into a workflow so everyone can see which have been solved, which need attention, and whos working on what. Google Docs and Google Sheets make it easy for multiple people to contribute to progress on the same puzzle whether theyre sitting side by side on campus or are separated by several time zones. I think especially post-2020, there is just the expectation that everything is going to be accessible online, says Tripathi, who still has a Hunt-related Google doc from 2008, just a couple of years after the service launched. But even as the Mystery Hunt has adapted to the internetand to increasingly powerful search engines, smartphones, the Zoom era, and even some machine-learning applicationsat its core it remains a very human experience. Its about connecting with other humansthats why we do it, says Erin Rhode 04, a longtime Mystery Hunter whose team has won twice. She recalls being inducted into the Hunt as a first-year in 2001. An upperclassman came in and was like, Youre coming to the math majors lounge. Were doing this puzzle hunt thing. The name of Rhodes team changes every year, though they might be best known for the year their name was the entire text of Ayn Rands Atlas Shrugged. Last year, they were . (Thats not a typo or a missing wordits the zero-width space, a Unicode non-character primarily used in document formatting.) Early Mystery Hunts led solvers to an Indian Head penny hidden on campus. Today, winning teams are awarded coins unique to each years Hunt. Ringed with a repeating MH24, the 2024 coin shows the cities teams visited on their quest.JADE CHONGSATHAPORNPONG 24/MIT TECHNIQUE Like so much of the Hunt, team names are an exercise in creativity. The full name of the team running the 2024 Mystery Hunt was officially The Team Formerly Known as the Team Formerly Known as the Team Formerly Known as the Team Formerly Known as the Team Formerly Known as the Team to Be Named Later. Some teams keep their name every year, like Setec Astronomy (an anagram for too many secrets, in a reference to the classic 1992 heist film Sneakers). Others change every year or every few years, or when teams merge, as when Death from Above joined forces with Project Electric Mayhem to become Death and Mayhem. Rhode remembers one particular puzzle from her first Hunt that she and her team (known that year as the Vermicious Knids) worked on through the night. They had to figure out that a list of enigmatic phrases were clues to song titles. For example, Of course; you just go north on Highway 101 clued the song Do You Know the Way to San Jose? I think today, we would have solved that puzzle in about an hour, Rhode says. There werent song lyric databases back then. And so it was a lot more sitting around on your own trying to come up with songs as opposed to just finding some master list and then searching it. Writing puzzles with the knowledge that solvers will have a slew of tools at hand is just part of the process. Use whatever technology you have at your disposal to solve the puzzle is the general rule of thumb, says Jon Schneider 13, a machine-learning researcher who hunts with Galactic Trendsetters . (The in their team name is pronounced like a plane taking off and landing, respectively.) Schneider has been hunting since 2010, when it was common for solvers to have to identify clips of songs or other audio. Hes seen that change in the past decade, though: Audio recognition [technology] like Shazam has become a thing, so its harder to create puzzles that require the skill of music recognition. When youre a constructor, you try to figure out: What is my challenge for the solver? says Dan Katz 03. Katz has solved and written a lot of puzzles. (In fact, he created a five-puzzle mini Hunt for this issues Puzzle Corner.) He attended his first Mystery Hunt in 1998, as a junior in high school, before he had even applied to MIT. Hes been part of a winning team eight times (probably a record) and competes in events like the World Sudoku Championship and US Puzzle Championship. In Katzs view, technology should make puzzling more interesting for the solver. While solvers might need to, say, code a program, organize information in a spreadsheet, or navigate a video-game-like interface to arrive at an answer, what he prizes most is the mental challenge of figuring out how to solve a puzzle. During whats known as the Mid-Hunt Runaround, a team follows a set of cryptic instructions that lead them on a subterranean journey across campus.JADE CHONGSATHAPORNPONG 24/MIT TECHNIQUE Rhode misses the days before an app was able to listen to a few seconds of a song and identify it. One of my superpowers in the early days of the Hunt was: Play me a bunch of pop songs and I can identify like 90% of them, she says. Now everybodys got Shazam on their phone. And so as fast as I might be, Shazam was always going to be faster. That doesnt mean puzzles cant be based on song identificationor image identification, another common puzzle element that has been made trivial by tools like Googles image search capabilities. It just means constructors must become more creative. You have to obscure the images or the music in such a way that the technology cant find it quickly, Rhode says. She describes a puzzle she wrote when she wanted solvers to identify songs without using technology: I arranged eight songs a cappella and sang them myself, but buzzing like a bee. And the whole idea was you cant Shazam that. Schneiders team took a similar approach to constructing a puzzle in which solvers had to identify specific visual artistsnot by their work, but by their distinctive style. Solvers were prompted to upload an image of their choosing, and a generative AI tool similar to DALL-E rendered it in the style of the artist they were supposed to name. I mostly justwant to be surprised. Jon Schneider 13 of the team Galactic Trendsetters Thats not the only puzzle to have incorporated some machine-learning elements in the last few years. A few examples have used semantic similarity scoring systems where solvers have to guess words or phrasesa kind of machine-learning-enabled version of hot or cold. Even if machine learning has potential as a tool for puzzle constructors, generative AI is unlikely to solve Mystery Hunt puzzles anytime soon. ChatGPT can answer questions that might be helpful in getting started and maybe even help solve a crossword clue or two, but the puzzles are often so unusual that it doesnt know where to begin. When presented with them, it usually responds by stating that it would need more context or clues in order to proceed. Schneider did find ChatGPT very helpful, though, in solving a nonMystery Hunt puzzle about navigating the byzantine rules of the role-playing game Dungeons & Dragons, which he admits hes never played. A few years ago, there would have been no way around spending hours digging through the rulebooks and figuring out each step, but giving the puzzle to ChatGPT worked. It was really good at doing this. I guess it had trained on enough data of people playing Dungeons & Dragons that this was within its capabilities, he says. Schneider is optimistic that new technology will be integrated into Mystery Hunt in creative ways, expanding the scope of what puzzle constructors can come up with to entertain solvers. Ultimately, he says, I mostly just want to be surprised. As the sun sets on Sunday, Setec continues solving puzzles at a steady pace, but were also still unlocking new sections of the Hunta sign that were still some distance from the endgame, though rumors (but never spoilers) from friends on other teams suggest that a few teams might be closing in. As midnight rolls around theres still no announcement, and so we push on. Ultimately, the 2024 Hunt ends up running into Monday morning, one of only a handful of times its taken more than 60 hours to complete. The 2024 Mystery Hunt included what was called the Herc-U-Lease Scavenger Hunt. As part of the scavenger hunt, teams were asked to have as many members as possible look as identical as possible. Death and Mayhem realized that many members were wearing black T-shirts and decided to unify the look with paper hats fashioned from copies of The Tech someone found on campus.MOLLY FREY/DEATH & MAYHEM A little after 5 a.m., team Death and Mayhem solves the final puzzle to win the 2024 Mystery Huntand the responsibility of developing the 2025 Hunt, which kicks off on January 17. In the end, 266 teams have solved at least one of the 2024 Hunts 237 puzzles and Setec Astronomy has solved 174. (Teams typically care less about postgame rankings than about how many puzzles they get to before time runs out.) The Team Formerly Known as the Team to Be Named Later sends out an announcement that a wrap-up event, at which theyll give a full overview of the weekend and hand over the reins to Death and Mayhem, will begin at noon in 26-100. Because creating a Mystery Hunt is such a daunting task, Death and Mayhem got to work on this years within hours of winning, says James Douberley 13, who assumed the title of benevolent dictator to orchestrate and oversee the teams puzzle writing. The weight of expectation is not lost on Douberley and his teammates: This is a once-a-year event that holds a lot of meaning for many participants. The Mystery Hunt is about solving puzzles, but its also far more social and immersive than puzzle books and escape rooms. In 2024, nearly 2,000 people representing 91 teams showed up on campus to participateand another 2,450 or so signed up to puzzle from afar. All told, solvers included 52 faculty members, 278 students, and 950 alumni, ranging from recent graduates to those who got their degrees decades ago. For Chang, the Hunt is an opportunity to connect with the broader community, including alumni from her dorm whom she doesnt see often. This is the one time in the year that we get to all just be in one place together and do this thing that we love, she says. Its just a really great bonding experience. Shortly after solving the final puzzles in the 2024 MIT Mystery Hunt, members of Death & Mayhem received the custom coins awarded to the victors and posed for a photo with Aphrodite (of the Team Formerly Known as the Team to Be Named Later), who blew kisses in celebration.COURTESY OF DEATH & MAYHEM The MIT campus plays a special role in the Hunt. Maybe you have to use the walls of the List Visual Arts Center lobby as a grid for a logic puzzle, or find certain names on the memorial plaques in Lobby 10 whose first letters spell out an answer. But its not just that clues can be part of the physical spaceits that campus is the epicenter for the MIT spirit of creativity, inventiveness, and industriousness that makes the Mystery Hunt unique. People talk about New York being a character in movies, Katz says. I feel like MIT is a character in Mystery Hunt. For Douberley, the Mystery Hunt takes him back to his student days, when he tackled hard challenges through marathon work sessions and all-nighters. You fall asleep on the floor, and youre in the dorm lounge and your friend comes and wakes you up and says, Heres a coffeeI need your help with something, he says. And that is something that lives with you for the rest of your life. Editors Note: The 2025 MIT Mystery Hunt kicks off on January 17, 2025. But if youre eager to start puzzling before thenor get a taste of puzzling if youve never taken part beforecheck out theMIT Mystery Heist, a pre-Hunt round of puzzles written by the Mystery Hunt team known as the Providence Crime Syndication. Learn more and solve atmitmysteryheist.com.

When I last wrote to you in this magazine, I told you a bit about the MIT Collaboratives, an effort to spark new ideas and modes of inquiry and help the people of MIT solve global problems. Since then, weve launched the first collaborative, grounding it in the human-centered fields represented by our School of Humanities, Arts, and Social Sciences (SHASS). Were calling it the MIT Human Insight Collaborative, or MITHIC. In broad terms, MITHIC is an endorsement of the quality of our faculty in these fields and an expression of how deeply we value the scholarly and artistic practices that expand our understanding of the things that make us human. In a practical sense, its designed to help our scholars in human-centered disciplines go big. MITHIC will give them the resources to pursue their most innovative ideas within their discipline, create opportunities for them to collaborate with colleagues outside it, and enable them to explore fresh approaches to teaching our students. We celebrated the launch of MITHIC with a showcase of creative excellence. MIT faculty shared research that blends the humanistic with the technological, MIT students improvised on jazz saxophone, and in a keynote conversation, the acclaimed novelist Min Jin Lee talked about her dedication to putting the human at the center of her work. Our faculty are wonderfully energized by MITHIC, and more than 100 have already taken part in the collaboratives Meeting of the Minds events, organized to connect researchers across the Institute who work on similar topicsfrom cybersecurity to food security, climate simulations to the bioeconomy. There may never have been a more important time for society to make humane choices about new technologies. And Im thrilled that at MIT weve created a collaborative powered by human insight to support our scholars, students, explorers, and makers in shaping a future of technology in service to humanity.

Im moving to Boston in three weeks! At my high school graduation, I had just learned Id been accepted into the Interphase EDGE program, an incredible opportunity to acclimate to life at MIT before the 2022 school year began. I was glad to have that chance, since I faced a big change from life at home in Claremore, on the Cherokee Nation reservation in northeastern Oklahoma. Id been away on my own only once, on a fifth-grade trip to Space Camp in Huntsville, Alabama, where I first fell in love with aerospace engineering. It didnt take long to find community on campus. To my surprise, out of the dozen students at a welcome event for the Indigenous community, three grad students and an undergrad were in the aero-astro department. As a prospective Course 16 major and a FIRST Robotics alum, I was excited to discover that they planned to start a new team for the First Nations Launch (FNL) rocketry competition, a NASA Artemis Student Challenge. It was the perfect opportunity to merge my technical passion with my cultural roots. That first year, many people questioned the need for our team. MIT already has a Rocket Team, theyd say. But while most build teams are defined by the specific projects they work on, the product is just one aspect of the experience. Yes, Ive learned to design, build, launch, and safely recover a model rocket. But doing that alongside other Indigenous engineers on the team we call MIT Doya (, Cherokee for beaver) has taught me more than engineering skills. Beyond learning how to work with composites or design fins, Ive learned how to navigate classes and connect with professors. Ive learned about grad school. And Ive learned how to celebrate my Indigenous identity and honor my ancestors with my work. For instance, we often hold smudging ceremoniesburning sage to purify ourselves or our rocketsat our team meetings and competitions. Our team emphasizes universal consensus and buy-in on the technical side and pays attention to the success of each team member on a personal level. We call this gadugi () in Cherokee, or everyone helping each other. Ive also learned that embracing my culture can offer a better approach to engineering challenges. While many engineering settings foster top-down decision-making, our team tests and incorporates as many ideas as possible to engage everyone, emphasizing universal consensus and buy-in on the technical side while paying attention to the success of each team member on a personal level. We call this gadugi () in Cherokee, or everyone helping each other. And we find its led to better technical resultsand a better experience for everyone on the team. I feel incredibly fortunate to work closely with other Indigenous students on an engineering project we all deeply care about. Ive looked up to the senior members of the team, seeing in them proof of what an Indigenous student at MIT can be and accomplish. And Ive loved mentoring newer members, passing along what Ive learned to help them excel. Our launch weekends expand our community further, allowing us to work alongside inspiring Indigenous engineers from NASAs Jet Propulsion Lab and Blue Origin. Ive gotten to meet my heroes and seen that its possible to succeed as a Native American in aerospace engineering. In fact, my FNL experiences have already helped me secure an amazing internship. Last summerexactly a decade after setting my heart on aerospace engineering at Space CampI returned to Huntsville as a lunar payloads intern on the Mark I Lunar Lander at Blue Origin. Through the FNL team, Ive significantly advanced my technical skills. As our systems and simulations lead the first year, I integrated all the components of the physical design into a cohesive computer model with accuracy in both geometry and mass distribution. From that model, I can run simulated flights while adjusting for various launch conditions and trying out different motors. A small change on the ground can yield a big change in our final altitude, which must be within a specific rangeso this analysis drives the overall design. In our first year, our challenge was to re-create the design of a kit rocket while making it lighter by fabricating all the parts ourselves, primarily using hand-laid carbon fiber and fiberglass. We finished in second place and were named Rookie Team of the Year. For 202324, our challenge was to build a rocket large enough to carry a deployable drone, leading us to build an airframe 7.5 inches in diameter. We also had to design and fabricate the drones chassis to meet strict specifications: It had to fit inside the rocket on the launchpad, deploy at apogee (ours was 2,136 feet), unfold from a compact stowed configuration to 16 by 16 inches, descend by parachute to 500 feet, and then release the parachute for piloted navigation to a landing pad. To meet FAA requirements, two of our team members studied for and earned Part 107 remote pilot certificates so they could operate the drone. Since this new challenge required us to fabricate a rocket while also designing and building the drone, we broke up into two subteams to work on both in parallel. This approach required precise coordination between the subteams to ensure that everything would integrate well for the final launch. As team captain, I managed this coordination while staying involved on the technical side as systems and simulations lead and airframe lead. And as we worked our way through the project milestones from proposal through flight readiness review, we kept in mind that we needed both an operational drone and a safe flight to the right altitude to meet the challenge. In April our team traveled to Kenosha, Wisconsin, to put our rocket to the test. We loaded the parachutes and payload, blessing it with some medicine before sending our hard work into the sky. But when I went to load our motor, the motor mount fell off in my hand. We quickly proceeded to the range safety officer, who was able to salvage our rocket and our launch with the last-minute addition of an external motor retention device. After that minor (but almost catastrophic) delay, we had a safe launch and successful recoveryand earned the Next Step Award, a $15,000 grant to represent FNL in the University Student Launch Initiative, a NASA-hosted competition open to everyone, for the 202425 season. Six weeks later, when the overall competition winners were announced, we were thrilled to learn we had won the grand prize! Along with bragging rights, we won a VIP trip to Kennedy Space Center in August and got to walk through the iconic Vehicle Assembly Building, explore the shuttle landing strip, see Polaris Dawn on the launchpad, and watch a Starlink launch from the beach in the early morning hours. This year, Im honored to serve as team captain again, leading an expanded team as we tackle the challenges of the new Student Launch Initiative. Im already looking forward to May, when well launch the rocket well be perfecting between now and then. And to honor our Indigenous heritage and send it into the sky with good intentions, Ill make sure we smudge before flight. Hailey Polson 26, an aero-astro major and a citizen of the Cherokee Nation, is captain of MITs First Nations Launch team.

This is today's edition ofThe Download,our weekday newsletter that provides a daily dose of what's going on in the world of technology. Will we ever trust robots? The world might seem to be on the brink of a humanoid-robot heyday. New breakthroughs in artificial intelligence promise the type of capable, general-purpose robots previously seen only in science fictionrobots that can do things like assemble cars, care for patients, or tidy our homes, all without being given specialized instructions. Its an idea that has attracted an enormous amount of attention, capital, and optimism. Yet recent progress has arguably been more about style than substance. Advancements in AI have undoubtedly made robots easier to train, but they have yet to enable them to truly sense their surroundings, think of what to do next, and carry out those decisions in the way some viral videos might imply.But on the road to helping humanoid robots win our trust, one question looms larger than any other: How much will they be able to do on his own? And how much will they still rely on humans? Read the full story.James ODonnell This story is from the forthcoming magazine edition of MIT Technology Review, set to go live on January 6its all about the exciting breakthroughs happening in the world right now. If you dont already, subscribe to receive future copies. If youre interested in the future of robots, why not check out: + A skeptics guide to humanoid-robot videos. The right video can land a startup millions in investment and a devoted public following. But what do these videos really show? + Is robotics about to have its own ChatGPT moment? Read the full story.+ To be more useful, robots need to become lazier. Smarter data processing could make machines more helpful and energy-efficient in the real world. A good way to test this principle is robot soccer.The must-reads Ive combed the internet to find you todays most fun/important/scary/fascinating stories about technology. 1 Anduril is in talks to join forces with OpenAI and SpaceX The proposed consortium will bid for US government defense contracts in an attempt to disrupt the stranglehold of traditional suppliers. (FT $)+ Elon Musks DOGE project could encourage similar partnerships. (Reuters)+ We saw a demo of the new AI system powering Andurils vision for war. (MIT Technology Review)2 Robotaxi passengers are targets of a new kind of harassment Riders feeling unsafe are left without a human driver to intervene. (WP $)+ Whats next for robotaxis. (MIT Technology Review)3 This covid season is the most unpredictable yetDeaths are down. But that doesnt mean we should fully relax. (The Atlantic $)4 WhatsApp has won its legal case against NSO Group The messaging app claims its spyware exploited a bug to surveil users. (Reuters)+ The case has been five years in the making. (WP $)+ NSO Group argued it wasnt liable as its software was used to investigate crimes. (The Verge)5 Why Elon Musk is turning his attention to right-wing UK politics Hes looking beyond the White House to the more extreme end of British mainstream political parties. (The Guardian)+ How seriously should we take Elon Musk? (New Yorker $)+ Donald Trump reminded activists that hes President-elect, not Musk. (NBC News)+ But how useful Musk will continue to be for Trump remains to be seen. (The Atlantic $)6 YouTube is finally cracking down on egregious clickbaitThe platform has long rewarded the creators behind misleading videos. (NY Mag $) + Hated that video? YouTubes algorithm might push you another just like it. (MIT Technology Review)7 What happens when AI collides with crypto In the wake of the NFT boom, something even scammier is stirring. (The Information $)+ What happens to bitcoin now that skeptics have become believers? (The Atlantic $)+ Its still not clear how AI will affect the economy. (Bloomberg $)+ How to fine-tune AI for prosperity. (MIT Technology Review)8 Beware of AI scams over the holidaysFrom fraudulent text messages to sneakily targeted ads. (WSJ $) + Five ways criminals are using AI. (MIT Technology Review)9 The highs and lows of 2024s viral moments I wont be holding space for them. (The Guardian) 10 NASAs fastest probe is heading for the suns atmosphere Itll endure temperatures of over 2,500 degrees Fahrenheit on Christmas Eve. (Wired $)+ Itll be the closest any probe has ever come to the sun. (Engadget)Quote of the day "I don't hate these people. I just hate being in their stupid group." Jess, a participant in a group chat for aspiring musicians, tells Insider why leaving the group before the new year is a top priority. The big story How Indian health-care workers use WhatsApp to save pregnant women February 2023 Across India, an all-women cadre of 1 million community health-care workers are responsible for making public health care accessible to people from remote areas and marginalized communities. These workers counsel pregnant women and ensure they receive proper science-backed health care. Many are turning to WhatsApp as a means to combat the medical misinformation that is rampant across the country and to navigate sensitive medical situations, particularly regarding pregnancy. Their approach has surprisingly good results. Read the full story. Sanket Jain We can still have nice things A place for comfort, fun and distraction to brighten up your day. (Got any ideas? Drop me a line or tweet 'em at me.) + Back in 2009, Rage Against the Machine reached Christmas number one in the UK music charts. Heres what happened when they were asked to censor their lyrics live on the BBC.+ Ever wished more films were like Home Alone? Youre in luck.+ How to make the perfect latke.+ No one has ever seen a flying reindeer. But that doesnt mean its a theoretical impossibility.

The world might seem to be on the brink of a humanoid-robot heyday. New breakthroughs in artificial intelligence promise the type of capable, general-purpose robots previously seen only in science fictionrobots that can do things like assemble cars, care for patients, or tidy our homes, all without being given specialized instructions. Its an idea that has attracted an enormous amount of attention, capital, and optimism. Figure raised $675 million for its humanoid robot in 2024, less than two years after being founded. At a Tesla event this past October, the companys Optimus robots outshined the self-driving taxi that was meant to be the star of the show. Teslas CEO, Elon Musk, believes that these robots could somehow build a future where there is no poverty. One might think that supremely capable humanoids are just a few years away from populating our homes, war zones, workplaces, borders, schools, and hospitals to serve roles as varied as therapists, carpenters, home health aides, and soldiers. Yet recent progress has arguably been more about style than substance. Advancements in AI have undoubtedly made robots easier to train, but they have yet to enable them to truly sense their surroundings, think of what to do next, and carry out those decisions in the way some viral videos might imply. In many of these demonstrations (including Teslas), when a robot is pouring a drink or wiping down a counter, it is not acting autonomously, even if it appears to be. Instead, it is being controlled remotely by human operators, a technique roboticists refer to as teleoperation. The futuristic looks of such humanoids, which usually borrow from dystopian Hollywood sci-fi tropes like screens for faces, sharp eyes, and towering, metallic forms, suggest the robots are more capable than they often are. Im worried that were at peak hype, says Leila Takayama, a robotics expert and vice president of design and human-robot interaction at the warehouse robotics company Robust AI. Theres a bit of an arms waror humanoids warbetween all the big tech companies to flex and show that they can do more and they can do better. As a result, she says, any roboticist not working on a humanoid has to answer to investors as to why. We have to talk about them now, and we didnt have to a year ago, Takayama told me. Shariq Hashme, a former employee of both OpenAI and Scale AI, entered his robotics firm Prosper into this arms race in 2021. The company is developing a humanoid robot it calls Alfie to perform domestic tasks in homes, hospitals, and hotels. Prosper hopes to manufacture and sell Alfies for approximately $10,000 to $15,000 each. Why are we enamored with this idea of building a replica of ourselves? Guy Hoffman, associate professor, Cornell University In conceiving the design for Alfie, Hashme identified trustworthiness as the factor that should trump all other considerations, and the top challenge that needs to be overcome to see humanoids benefit society. Hashme believes one essential tactic to get people to put their trust in Alfie is to build a detailed character from the ground upsomething humanlike but not too human. This is about more than just Alfies appearance. Hashme and his colleagues are envisioning the way the robot moves and signals what hell do next; imagining desires and flaws that shape his approach to tasks; and crafting an internal code of ethics that governs the instructions he will and will not accept from his owners. In some ways, leaning so heavily on the principle of trustworthiness for Alfie feels premature; Prosper has raised a tiny amount of capital compared with giants like Tesla or Figure and is months (or years) away from shipping a product. But the need to tackle the issue of trustworthiness head-on and early reflects the messy moment humanoids are in: Despite all the investment and research, few people would feel warm and comfortable with such a robot if it walked into their living room right now. Wed wonder what data it was recording about us and our surroundings, fear it might someday take our job, or be turned off by its way of moving; rather than elegant and useful, humanoids are often cumbersome and creepy. Overcoming that lack of trust will be the first hurdle to clear before humanoids can live up to their hype. But on the road to helping Alfie win our trust, one question looms larger than any other: How much will he be able to do on his own? How much will he still rely on humans? New AI techniques have made it faster to train robots through demonstration datausually some combination of images, videos, and other data created by humans doing tasks like washing dishes while wearing sensors that pick up on their movements. This data can then coach robots through those tasks much the way that a large body of text can help a large language model create sentences. Still, this method requires lots of data, and lots of humans need to step in and correct for errors. Hashme told me that he expects the first release of Alfie to handle only about 20% of tasks on his own. The rest will be assisted by a Prosper team of remote assistants, at least some of them based in the Philippines, who will have the ability to remotely control Alfies movements. When I raised, among other concerns, whether its viable for a robotics business to rely on manual human labor for so many tasks, Hashme pointed to the successes of Scale AI. That company, which processes training data for AI applications, has a significant workforce in the Philippinesand is often criticized for its labor practices. Hashme was one of the people managing that workforce for about a year before founding Prosper. His departure from Scale AI was itself set off by a violation of trustone for which he would serve time in federal prison. The success or failure of Alfie will reveal much about societys willingness to welcome humanoid robots into our private spaces. Can we accept a profoundly new and asymmetric labor arrangement in which workers in low-wage countries use robotic interfaces to perform physical tasks for us at home? Will we trust them to safeguard private data and images of us and our families? On the most basic level, will the robots even be useful? To address some of these concerns around trust, Hashme brought in Buck Lewis. Two decades before Lewis worked with robots, before he was charged with designing a humanoid that people would trust rather than fear, the challenge in front of him was a rat. In 2001, Lewis was a revered animator and one of the top minds at Pixar. His specialty was designing characters with deep, universal appeal, a top concern to studios that fund high-budget projects aimed at capturing audiences worldwide. It was a niche that had led Lewis to bring trucks and sedans to life in the movie Cars and create characters for many DreamWorks and Disney films. But when Jan Pinkava, the creative force behind Ratatouille, told Lewis about his pitch for that filmthe story of a rat who wants to be a chefthe task felt insurmountable. Rats evoke such fear and apprehension in humans that their very name has become a shorthand for someone who cannot be trusted. How could Lewis turn a maligned rodent into an endearing chef? Its a deeply ingrained aversion, because rats are horrifying, he told me. For this to work, we had to create a character that rewires peoples perceptions. To do that, Lewis spent a lot of time in his head, imagining scenes like a group of rats hosting a playful pop-up dinner on a sidewalk in Paris. The result was Remy, a Parisian rat who not only rose through the culinary ranks in Ratatouille but was so lovable that demand for pet rats surged globally after the films release in 2007. Two decades later, Lewis has made a career change and is now in charge of crafting every aspect of Alfies character at Prosper. Much as the appealing Remy rebranded rats, Alfie represents Lewiss attempt to change the image of humanoid robots, from futuristic and dangerous to helpful and trustworthy. Prospers approach reflects a foundational robotics concept articulated by Rodney Brooks, a founder of iRobot, which created the Roomba: The visual appearance of a robot makes a promise about what it can do and how smart it is. It needs to deliver or slightly overdeliver on that promise or it will not be accepted. According to this principle, any humanoid robot makes the promise that it can behave like a humanwhich is an exceedingly high bar. So high, in fact, that some firms reject it. Some humanoid-skeptic roboticists doubt that a helpful robot needs to resemble a human at all when it could instead accomplish practical tasks without being anthropomorphized. Why are we enamored with this idea of building a replica of ourselves? asks Guy Hoffman, a roboticist focused on human-robot interactions and an associate professor at Cornell Universitys engineering school. Early prototypes of Prospers robotic butler, which could perform household tasks like cleaning a kitchen table, rinsing dishes, and discarding trash.DAVID VINTINER The chief argument for robots with human characteristics is a functional one: Our homes and workplaces were built by and for humans, so a robot with a humanlike form will navigate them more easily. But Hoffman believes theres another reason: Through this kind of humanoid design, we are selling a story about this robot that it is in some way equivalent to us or to the things that we can do. In other words, build a robot that looks like a human, and people will assume its as capable as one. In designing Alfies physical appearance, Prosper has borrowed some aspects of typical humanoid design but rejected others. Alfie has wheels instead of legs, for example, as bipedal robots are currently less stable in home environments, but he does have arms and a head. The robot will be built on a vertical column that resembles a torso; his specific height and weight are not yet public. He will have two emergency stop buttons. Nothing about Alfies design will attempt to obscure the fact that he is a robot, Lewis says. The antithesis [of trustworthiness] would be designing a robot thats intended to emulate a human and its measure of success is based on how well it has deceived you, he told me. Like, Wow, I was talking to that thing for five minutes and I didnt realize its a robot. That, to me, is dishonest. But much other humanoid innovation is headed in a direction where deception seems to be an increasingly attractive concept. In 2023, several ultrarealistic humanoid robots appeared in the crowd at an NFL game at SoFi stadium in California; after a video of them went viral, Disney revealed they were actually just people in suits, a stunt to promote a movie. Nine months later, researchers from the University of Tokyo unveiled a way to attach engineered skin, which used human cells, over the face of a robot in an attempt to more perfectly resemble a human face. Through this kind of humanoid design, we are selling a story [that this robot] is in some way equivalent to us or to the things that we can do. Guy Hoffman, roboticist Lewis has considered much more than just Alfies appearance. He and Prosper envision Alfie as an ambassador from a future civilization in which robots have incorporated the best qualities of humanity. Hes not young or old but has the wisdom of middle age, and his primary function in life is to be of service to people on their terms. Like any compelling character, Alfie has flaws people can relate tohe wishes he could be faster, and he tends to be a bit obsessive about finishing the tasks asked of him. Core tenets of Alfies service are to respect boundaries, to be discreet and nonjudgmental, and to earn trust. Hes an entity thats nonhuman, but he has a sort of sentience, Lewis says. Im trying to avoid looking at it as directly comparable to human consciousness. Ive been referring to Alfie as heat the risk of over-anthropomorphizing what is currently a robot in developmentbecause Lewis pictures him as a gendered male. When I asked why he pictures Alfie as having a gender, he said its probably a relic from the archetypal male butlers he saw on television shows like Batman growing up. But in a conversation with Hashme, I learned there is actually a real-life butler who is in some ways serving as an inspiration for Alfie. That would be Fitzgerald Heslop. Heslop has decades of experience in high-end hospitality training, and for seven years he was the only person within the United States Department of Defense qualified to train household managers who would run the homes of three- and four-star generals. Heslop now runs the household of a wealthy family in the Middle East (he declined to get more specific) and has been contracted by Prosper to inform Alfies approach to service within the home. Shortly into my conversation with Heslop, he elaborated on what excellent service looks like. Thats the level of creativity the good butler deals in: the making of beautiful moments to put people at their ease and increase their pleasure, he said, quoting Steven M. Ferrys book Butlers & Household Managers: 21st Century Professionals. He spoke with conviction about the impact great service can have on the world and about how protocol and etiquette can level the egos of even top dignitaries. Citing a quote often attributed to Mahatma Gandhi, he said, The best way to find yourself is to lose yourself in the service of others. Though he has no experience in robotics, Heslop is drawn to the idea that household robots could someday provide impeccable levels of service, and he thinks that Prosper has identified the right priorities to get there. Privacy and discretion, attention to detail, and meticulous eyes for that are mission critical to the overall objective of the company, he says. And more importantly, in this case, Alfie. It is one thing to dream up an Alfie in sketchbooks, and another to build him. In the real world, the first version of Alfie will depend on remote assistants, mostly working abroad, to handle approximately 80% of its household tasks. These assistants will use interfaces not unlike video-game controllers to control Alfies movements, relying on data from his sensors and cameras to guide them in washing dishes or clearing a table. Hashme says efforts are being made to conceal or anonymize personally revealing data while the robot is being teleoperated. That will include steps like removing sensitive objects and peoples faces from recordings and allowing users to delete any footage they like. Ideally, Hashme wrote in an email, Alfie will often simply look away from any potentially private activities. The AI industry has an appalling track record when it comes to workers in low-wage countries performing the hidden labor required to build cutting-edge models. Workers in Kenya were reportedly paid less than $2 an hour to manually purge toxic training data, including content describing child sexual abuse and torture, for OpenAI. Scale AIs own operation in the Philippines, which Hashme helped manage, was criticized in 2023 by rights groups for not abiding by basic labor standards and failing to pay workers properly and on time, according to an investigation by the Washington Post. In a statement, OpenAI said such work needs to be done humanely and willingly, and that the company establishes ethical and wellness standards for our data annotators. In a response to questions about criticisms of its operation in the Philippines, Scale AI wrote, Over the past year alone, weve paid out hundreds of millions in earnings to contributors, giving people flexible work options and economic opportunity, and that 98% of support tickets regarding pay have been successfully resolved. Hashme says he was not aware of the allegations against ScaleAI during his time there, which ended in 2019. But, he said in an email, we did make mistakes, which we quickly corrected and generally took quite seriously. I asked him what lessons he takes from the allegations against Scale AI and other companies outsourcing sensitive data work and what safeguards hes putting in place for the team hes building in the Philippines for Prosper, which so far numbers about 10 people. Shariq Hashme, a former employee of both OpenAI and Scale AI, entered his robotics firm Prosper into the humanoid arms race in 2021.DAVID VINTINER A lot of companies that do that kind of stuff end up doing it in a way which is kind of shitty for the people who are being employed, Hashme told me. Such companies often outsource important HR activities to untrustworthy partners abroad or lose workers trust through bad incentive programs, he said, adding: With a more experienced and closely managed team, and a lot more transparency around the entire system, I expect well be able to do a much better job. Its worth disclosing the nature of Hashmes departure from Scale AI, where he was hired in 2017 as its 14th employee. In May 2019, according to court documents, Scale noticed that someone had repeatedly withdrawn unauthorized payments of $140 and transferred them to multiple PayPal accounts. The company contacted the FBI. Over the course of five months, approximately $56,000 was taken from the company. An investigation revealed that Hashme, then 26, was behind the withdrawals, and in October of that year, he pleaded guilty to one count of wire fraud.Ahead of his sentencing, Alexandr Wang, the now-billionaire founder and CEO of Scale AI, wrote a letter to the judge in support of Hashme, as did 13 other current or former Scale employees. I believe Shariq is genuinely remorseful for his crime, and I have no reason to believe he will ever do something like this again, Wang wrote, and he said the company would not have wanted the wrongdoer prosecuted if it had known it was Hashme. Hashme lost his job, his stock options, and Scales sponsorship of his green card application. Scale offered him a $10,000 severance payment before leaving, which he declined to accept, according to Wangs letter. Hashme paid the money back in 2019, and in February 2020, he was sentenced to three months in federal prison, which he served. Wang is now a primary investor in Prosper Robotics, alongside Ben Mann (cofounder of Anthropic), Simon Last (cofounder of Notion), and Debo Olaosebikan (cofounder and CEO of Kepler Computing). I had a major lapse in judgment when I was younger. I was facing some personal challenges and stole from my employer. The consequences and the realization of what Id done came as a shock, and led to a lot of soul-searching, Hashme wrote in an email in response to questions about the crime. At Prosper, he wrote, were taking trustworthiness as our highest aspiration. There are some real upsides to being able to control robots remotely, but the idea of large-scale robotic teleoperation by overseas workers, even if it takes years for it to be effective, would be nothing short of a seismic shift for labor. It would present the possibility that even highly localized physical work that we perceive as immune to moving offshorecleaning hotel rooms or caring for hospital patientsmight someday be conducted by workers abroad. It also seems antithetical to the very idea of a trustworthy robot, since the machines effectiveness would be inextricably tied to a faceless worker in another country, most likely receiving paltry wages. Hashme has spoken about using a portion of Prospers profits to make direct payments to people whose jobs have been affected or replaced by Alfies, but he doesnt have specifics on how that would work. Hes also still thinking through issues related to who or what Prospers customers should be trusting when they allow its robot into their home. We dont want you to have to place as much trust in the company or the people the company hires, he says. Wed rather you place trust in the device, and the device is the robot, and the robot is making sure the company doesnt do something theyre not supposed to do. He admits that the first version of Alfie will likely not live up to his highest aspirations, but he remains steadfast that the robot can be of service to society and to people, if only they can trust him.

Its a stormy holiday weekend, and youve just received the last notification you want in the busiest travel week of the year: The first leg of your flight is significantly delayed. You might expect this means youll be sitting on hold with airline customer service for half an hour. But this time, the process looks a little different: You have a brief text exchange with the airlines AI chatbot, which quickly assesses your situation and places you in a priority queue. Shortly after, a human agent takes over, confirms the details, and gets you rebooked on an earlier flight so you can make your connection. Youll be home in time to enjoy moms pot roast. DOWNLOAD THE REPORT Generative AI is becoming a key component of business operations and customer service interactions today. According to Salesforce research, three out of five workers (61%) either currently use or plan to use generative AI in their roles. A full 68% of these employees are confident that the technologywhich can churn out text, video, image, and audio content almost instantaneouslywill enable them to provide more enriching customer experiences. But the technology isnt a complete solutionor a replacement for human workers. Sixty percent of the surveyed employees believe that human oversight is indispensable for effective and trustworthy generative AI. Generative AI enables people and increases efficiencies in business operations, but using it to empower employees will make all the difference. Its full business value will only be achieved when it is used thoughtfully to blend with human empathy, ingenuity, and emotional intelligence. Generative AI pilots across industries Though the technology is still nascent, many generative AI use cases are starting to emerge. In sales and marketing, generative AI can assist with creating targeted ad content, identifying leads, upselling, cross-selling, and providing real-time sales analytics. When used for internal functions like IT, HR, and finance, generative AI can improve help-desk services, simplify recruitment processes, generate job descriptions, assist with onboarding and exit processes, and even write code. Download the full report. This content was produced by Insights, the custom content arm of MIT Technology Review. It was not written by MIT Technology Reviews editorial staff.

A live agent spends hours each week manually documenting routine interactions. Another combs through multiple knowledge bases to find the right solution, scrambling to piece it together while the customer waits on hold. A third types out the same response theyve written dozens of times before. These repetitive tasks can be draining, leaving less time for meaningful customer interactionsbut generative AI is changing this reality. By automating routine workflows, AI augments the efforts of live agents, freeing them to do what they do best: solving complex problems and applying human understanding and empathy to help customers during critical situations. DOWNLOAD THE REPORT Enterprises are trying to rush to figure out how to implement or incorporate generative AI into their business to gain efficiencies, says Will Fritcher, deputy chief client officer at TP. But instead of viewing AI as a way to reduce expenses, they should really be looking at it through the lens of enhancing the customer experience and driving value. Doing this requires solving two intertwined challenges: empowering live agents by automating routine tasks and ensuring AI outputs remain accurate, reliable, and precise. And the key to both these goals? Striking the right balance between technological innovation and human judgment. A key role in customer support Generative AIs potential impact on customer support is twofold: Customers stand to benefit from faster, more consistent service for simple requests, while also receiving undivided human attention for complex, emotionally charged situations. For employees, eliminating repetitive tasks boosts job satisfaction and reduces burnout.The tech can also be used to streamline customer support workflows and enhance service quality in various ways, including:Automated routine inquiries: AI systems handle straightforward customer requests, like resetting passwords or checking account balances. Real-time assistance: During interactions, AI pulls up contextually relevant resources, suggests responses, and guides live agents to solutions faster. Fritcher notes that TP is relying on many of these capabilities in its customer support solutions. For instance, AI-powered coaching marries AI-driven metrics with human expertise to provide feedback on 100% of customer interactions, rather than the traditional 2% to 4% that was monitored pre-generative AI.Call summaries: By automatically documenting customer interactions, AI saves live agents valuable time that can be reinvested in customer care. Download the full report. This content was produced by Insights, the custom content arm of MIT Technology Review. It was not written by MIT Technology Reviews editorial staff.

David Dewan came up with this solution (which Richard Lipesalso found on Wikipedia): Let N be any integer not divisible by 2 or 5. Consider repunits R1 = 1, R2 = 11, R3 = 111, , RN+1 and their residues modulo N. There are at most N different residues, so the set of N+1 residues modulo N must contain at least one repeat. Assume RA mod N and RB mod N with B>A are the same. Then (RB RA) mod N = (RB-A10A) mod N = 0. Since N and 10 are relatively prime, N divides RB-A. S/O4. Frank offers this sudoku problem: Many readers tackled it successfully. For anyone who got stuck, heres the answer key: Ten years ago I was doing what is now called AI, Richard Marks 58 writes. He noted that trying all possible iterations of a sudoku problem will tie up your Cray for a week. So I personally wrote the rules and coded this little AI program that solves any sudoko by the time you have released the Solve It button. Since I wrote the program from scratch, I guess you can say I solved S/O4. S/O6. On behalf of the MIT Chess Club, Justin Zhou 25 asked how White can play and mate in two (see below). Frank Model 63 says there are two cases to consider: when Black can castle, and when Black cannot. If Black can castle, then Blacks last move had to be Pc7-c5. In this case White can take en passant, Pb5-c6. If Black castles, then White plays Pb7#. If Black can castle, but makes some other move, then White plays Rf8#. If Black cannot castle, then White plays Ke6. Black cannot escape mate on the next move when White plays Rf8#. Steve Gordon noted that this problem illustrates the three special chess moves. The following was adapted from his analysis with his algebraic chess notation. Black last moved either a king, rook, or c5. If c5, then 1. bxc6 (en passant a.k.a e. p.). If Black can still castle queenside (1. Kc8 & Rd8, a.k.a O-O-O), then 2. b7#. If Black cannot, its king is still trapped on rank 8, so after any black move, 2. Rf8#. If Black last moved a king or rook, en passant is not possible for White, but Black cant castle either, so 1. Ke6 also traps the black king on rank 8, and after any Black move, 2. Rf8#. Note: To create this puzzle, the white bishop on g7 resulted from an underpromotion on h8.

Ready for a fresh set of puzzles? Click here for the January/February Puzzle Corner, brought to you with a special Mystery Hunt twist by guest editor Dan Katz 03.This column includes solutions to three September/October 24 problems. Find solutions to the other three problems here.

This is today's edition ofThe Download,our weekday newsletter that provides a daily dose of what's going on in the world of technology. The next generation of neural networks could live in hardware Networks programmed directly into computer chip hardware can identify images faster, and use much less energy, than the traditional neural networks that underpin most modern AI systems. Thats according to work presented at a leading machine learning conference in Vancouver last week. Neural networks, from GPT-4 to Stable Diffusion, are built by wiring together perceptrons, which are highly simplified simulations of the neurons in our brains. In very large numbers, perceptrons are powerful, but they also consume enormous volumes of energy. Part of the trouble is that perceptrons are just software abstractionsrunning a perceptron network on a GPU requires translating that network into the language of hardware, which takes time and energy. Building a network directly from hardware components does away with a lot of those costs. And one day, they could even be built directly into chips used in smartphones and other devices. Read the full story. Grace Huckins Drugs like Ozempic now make up 5% of prescriptions in the US Whats new? US doctors write billions of prescriptions each year. During 2024, though, one type of drug stood outwonder drugs known as GLP-1 agonists. As of September, one of every 20 prescriptions written for adults was for one of these drugs, according to the health data company Truveta. The big picture: According to the data, people who get prescriptions for these drugs are younger, whiter, and more likely to be female. In fact, women are twice as likely as men to get a prescription. Yet not everyone whos prescribed the drugs ends up taking them. In fact, half the new prescriptions for obesity are going unfilled. Read the full story. Antonio Regalado Why childhood vaccines are a public health success story Childhood vaccination is a success story. In the 50 years since the World Health Organization launched its ambitious global childhood vaccination program, vaccines are estimated to have averted 154 million deaths. That number includes 146 million children under the age of five. But concerns around vaccines endure. Especially, it seems, among the individuals Donald Trump has picked as his choices to lead US health agencies from January. So lets take a look at their claims, and where the evidence really stands on childhood vaccines. Read the full story. Jessica Hamzelou This story is from The Checkup, our weekly health and biotech newsletter. Sign up to receive it in your inbox every Thursday. The must-reads Ive combed the internet to find you todays most fun/important/scary/fascinating stories about technology. 1 Elon Musk is the shadow president of the United States The billionaire pressured Republicans into impeding a spending bill, despite lacking an official government role. (WP $)+ He posted about the bill more than 100 times on Wednesday alone (NBC News)+ but those posts were generally misleading or outright false. (Rolling Stone $)+ Lawmakers arent thrilled about Musks interference. (NYT $)2 Amazon workers are striking during the Christmas rush The walkouts could delay the delivery of parcels across the US. (WSJ $)+ Amazon is refusing to recognize the workers labor union. (WP $)3 The US is growing increasingly wary of Nvidias overseas sales spree Officials worry the chipmakers deals could end up empowering its adversaries. (NYT $)+ US-based venture firms have pledged to avoid taking funding from China. (WP $)+ Custom chipmaker Broadcoms stock is surging right now. (Insider $)4 Dozens of families are suing Snap over teen overdoses They allege Snapchat helped dealers to sell deadly counterfeit drugs to their children. (Bloomberg $)5 Ukraines drone footage will be used to train AI models The country has collected 228 years worth of data during its conflict with Russia. (Reuters)+ An overnight drone attack set fire to a refinery in south Russia. (Bloomberg $)+ Meet the radio-obsessed civilian shaping Ukraines drone defense. (MIT Technology Review)6 Jailbreaking AI models can be as simple as TyPiNg LiKe ThIsAnd the methods are simple to automate, too. (404 Media) + Text-to-image AI models can be tricked into generating disturbing images. (MIT Technology Review) 7 Indias answer to Silicon Valley is under immense pressureBengalurus rapid expansion is pushing the citys infrastructure to the absolute limit. (Insider $) + Indias gig economy is focusing on 10-minute deliveries. (Bloomberg $)+ How Indian health-care workers use WhatsApp to save pregnant women. (MIT Technology Review)8 Whats next for AI gadgets?Consumers werent overly enamored with them in 2024. (Fast Company $) 9 The man who claimed to have created bitcoin has been sentenced Craig Wright has been given a one-year suspended sentence after refusing to stop suing developers. (The Guardian)+ Hell face jail if he continues claiming he really is the mysterious Satoshi Nakamoto. (BBC)10 Online returns arent what they used to be Retailers are fed up, and so are customers. (The Atlantic $)Quote of the day You guys scared the life out of a lot of people. Geno, an Arizona resident, tells Amazon workers that their delivery drones are making his neighbors uneasy amid the drone panic gripping the US, the New York Times reports. The big story Bright LEDs could spell the end of dark skies August 2022 Scientists have known for years that light pollution is growing and can harm both humans and wildlife. In people, increased exposure to light at night disrupts sleep cycles and has been linked to cancer and cardiovascular disease, while wildlife suffers from interruption to their reproductive patterns, and increased danger. Astronomers, policymakers, and lighting professionals are all working to find ways to reduce light pollution. Many of them advocate installing light-emitting diodes, or LEDs, in outdoor fixtures such as city streetlights, mainly for their ability to direct light to a targeted area. But the high initial investment and durability of modern LEDs mean cities need to get the transition right the first time or potentially face decades of consequences. Read the full story. Shel Evergreen We can still have nice things A place for comfort, fun and distraction to brighten up your day. (Got any ideas? Drop me a line or skeet 'em at me.) + How the Black diaspora celebrates Christmas across the world, featuring Motown tunes and a tasty saltfish salad.+ We love you Pamela Anderson!+ Test your science knowledge with this fiendish quiz of the year.+ Lets look ahead to just some of the exciting films coming out next year, from Bridget Jones to the bonkers-sounding Mickey 17.

This article first appeared in The Checkup,MIT Technology Reviewsweekly biotech newsletter. To receive it in your inbox every Thursday, and read articles like this first,sign up here. Later today, around 10 minutes after this email lands in your inbox, Ill be holding my four-year-old daughter tight as she receives her booster dose of the MMR vaccine. This shot should protect her from a trio of nasty infectionsinfections that can lead to meningitis, blindness, and hearing loss. I feel lucky to be offered it. This year marks the 50-year anniversary of an ambitious global childhood vaccination program. The Expanded Programme on Immunization was launched by the World Health Organization in 1974 with the goal of getting lifesaving vaccines to all the children on the planet. Vaccines are estimated to have averted 154 million deaths since the launch of the EPI. That number includes 146 million children under the age of five. Vaccination efforts are estimated to have reduced infant mortality by 40%, and to have contributed an extra 10 billion years of healthy life among the global population. Childhood vaccination is a success story. But concerns around vaccines endure. Especially, it seems, among the individuals Donald Trump has picked as his choices to lead US health agencies from January. This week, lets take a look at their claims, and where the evidence really stands on childhood vaccines. WHO, along with health agencies around the world, recommends a suite of vaccinations for babies and young children. Some, such as the BCG vaccine, which offers some protection against tuberculosis, are recommended from birth. Others, like the vaccines for pertussis, diphtheria, tetanus, and whooping cough, which are often administered in a single shot, are introduced at eight weeks. Other vaccinations and booster doses follow. The idea is to protect babies as soon as possible, says Kaja Abbas of the London School of Hygiene & Tropical Medicine in the UK and Nagasaki University in Japan. The full vaccine schedule will depend on what infections pose the greatest risks and will vary by country. In the US, the recommended schedule is determined by the Centers for Disease Control and Prevention, and individual states can opt to set vaccine mandates or allow various exemptions. Some scientists are concerned about how these rules might change in January, when Donald Trump makes his return to the White House. Trump has already listed his picks for top government officials, including those meant to lead the countrys health agencies. These individuals must be confirmed by the Senate before they can assume these roles, but it appears that Trump intends to surround himself with vaccine skeptics. For starters, Trump has selected Robert F. Kennedy Jr. as his pick to lead the Department of Health and Human Services. Kennedy, who has long been a prominent anti-vaxxer, has a track record of spreading false information about vaccines. In 2005, he published an error-laden article in Salon and Rolling Stone linking thimerosalan antifungal preservative that was previously used in vaccines but phased out in the US by 2001to neurological disorders in children. (That article was eventually deleted in 2011. I regret we didnt move on this more quickly, as evidence continued to emerge debunking the vaccines and autism link, wrote Joan Walsh, Salons editor at large at the time.) Kennedy hasnt let up since. In 2015, he made outrageous comments about childhood vaccinations at a screening of a film that linked thimerosal to autism. They get the shot, that night they have a fever of a hundred and three, they go to sleep, and three months later their brain is gone, Kennedy said, as reported by the Sacramento Bee. This is a holocaust, what this is doing to our country. Aaron Siri, the lawyer who has been helping Kennedy pick health officials for the upcoming Trump administration, has petitioned the government to pause the distribution of multiple vaccines and to revoke approval of the polio vaccine entirely. And Dave Weldon, Trumps pick to direct the CDC, also has a history of vaccine skepticism. He has championed the disproven link between thimerosal and autism. These arguments arent new. The MMR vaccine in particular has been subject to debate, controversy, and conspiracy theories for decades. All the way back in 1998, a British doctor, Andrew Wakefield, published a paper suggesting a link between the vaccine and autism in children. The study has since been debunkedmultiple times overand Wakefield was found to have unethically subjected children to invasive and unnecessary procedures. The paper was retracted 12 years after it was published, and the UKs General Medical Council found Wakefield guilty of serious professional misconduct. He was struck off the medical register and is no longer allowed to practice medicine in the UK. (He continues to peddle false information, though, and directed the 2016 film Vaxxed, which Weldon appeared in.) So its remarkable that his study still seems to be affecting public opinion. A recent Pew Research Center survey suggests that four in 10 US adults worry that not all vaccines are necessary, and while most Americans think the benefits outweigh any risks, some are still concerned about side effects. Views among Republicans in particular seem to have shifted over the years. In 2019, 82% supported school-based vaccine requirements. That figure dropped to 70% in 2023. The problem is that we need more than 70% of children to be vaccinated to reach herd immunitythe level needed to protect communities. For a super-contagious infection like measles, 95% of the population needs to be vaccinated, according to WHO. If [coverage drops to] 80%, we should expect outbreaks, says Abbas. And thats exactly what is happening. In 2023, only 83% of children got their first dose of a measles vaccine through routine health services. Nearly 35 million children are thought to have either partial protection from the disease or none at all. And over the last five years, there have been measles outbreaks in 103 countries. Polio vaccinesthe ones whose approval Siri sought to revokehave also played a vital role in protecting children, in this case from a devastating infection that can cause paralysis. People were so afraid of polio in the 30s, 40s, and 50s here in the United States, says William Moss, an epidemiologist at Johns Hopkins Bloomberg School of Public Health in Baltimore, Maryland. When the trial results of [the first] vaccine were announced in the United States, people were dancing in the streets. That vaccine was licensed in the US in 1955. By 1994, polio was considered eliminated in North and South America. Today, wild forms of the virus have been eradicated in all but two countries. But the polio vaccine story is not straightforward. There are two types of polio vaccine: an injected type that includes a dead form of the virus, and an oral version that includes live virus. This virus can be shed in feces, and in places with poor sanitation, it can spread. It can also undergo genetic changes to create a form of the virus that can cause paralysis. Although this is rare, it does happenand today there are more cases of vaccine-derived polio than wild-type polio. It is worth noting that since 2000, more than 10 billion doses of the oral polio vaccine have been administered to almost 3 billion children. It is estimated that more than 13 million cases of polio have been prevented through these efforts. But there have been just under 760 cases of vaccine-derived polio. We could prevent these cases by switching to the injected vaccine, which wealthy countries have already done. But thats not easy in countries with fewer resources and those trying to reach children in remote rural areas or war zones. Even the MMR vaccine is not entirely risk-free. Some people will experience minor side effects, and severe allergic reactions, while rare, can occur. And neither vaccine offers 100% protection against disease. No vaccine does. Even if you vaccinate 100% [of the population], I dont think well be able to attain herd immunity for polio, says Abbas. Its important to acknowledge these limitations. While there are some small risks, though, they are far outweighed by the millions of lives being saved. [People] often underestimate the risk of the disease and overestimate the risk of the vaccine, says Moss. In some ways, vaccines have become a victim of their own success. Most of todays parents fortunately have never seen the tragedy caused by vaccine-preventable diseases such as measles encephalitis, congenital rubella syndrome, and individuals crippled by polio, says Kimberly Thompson, president of Kid Risk, a nonprofit that conducts research on health risks to children. With some individuals benefiting from the propagation of scary messages about vaccines and the proliferation of social media providing reinforcement, its no surprise that fears may endure. But most Americans recognize the benefits of vaccines and choose to get their children immunized, she adds. Now, that is a sentiment I can relate to. Now read the rest of The Checkup Read more from MIT Technology Review's archive A couple of years ago, the polio virus was detected in wastewater in London, where I live. I immediately got my daughter (who was only one year old then!) vaccinated. Measles outbreaks continue to spring up in places where vaccination rates drop. Researchers hope that searching for traces of the virus in wastewater could help them develop early warning systems. Last year, the researchers whose work paved the way for the development of mRNA vaccines were awarded the Nobel Prize. Now, scientists are hoping to use the same technology to treat and vaccinate against a host of diseases. Most vaccines work by priming the immune system to respond to a pathogen. Scientists are also working on inverse vaccines that teach the immune system to stand down. They might help treat autoimmune disorders. From around the web A person in the US is the first in the country to have become severely ill after being infected with the bird flu virus, the US Centers for Disease Control and Prevention shared on December 18. The case was confirmed on December 13. The person was exposed to sick and dead birds in backyard flocks in Louisiana. (CDC) Gavin Newsom, the governor of California, declared a state of emergency as the bird flu virus moved from the Central Valley to Southern California dairy herds. Since August, 645 herds have been reported to be infected with the virus. (LA Times) Pharmacy benefit managers control access to prescription drugs for most Americans. These middlemen were paid billions of dollars by drug companies to allow the free flow of opioids during the USs deadly addiction epidemic, an investigation has revealed. (New York Times) Weight-loss drugs like Ozempic have emerged as blockbuster medicines over the past couple of years. Were learning that they may have benefits beyond weight loss. Might they also protect organ function or treat kidney disease? (Nature Medicine) Doctors and scientists have been attempting head transplants on animals for decades. Can they do it in people? Watch this delightful cartoon to learn more about the early head transplant attempts. (Aeon)

Networks programmed directly into computer chip hardware can identify images faster, and use much less energy, than the traditional neural networks that underpin most modern AI systems. Thats according to work presented at a leading machine learning conference in Vancouver last week. Neural networks, from GPT-4 to Stable Diffusion, are built by wiring together perceptrons, which are highly simplified simulations of the neurons in our brains. In very large numbers, perceptrons are powerful, but they also consume enormous volumes of energyso much that Microsoft has penned a deal that will reopen Three Mile Island to power its AI advancements. Part of the trouble is that perceptrons are just software abstractionsrunning a perceptron network on a GPU requires translating that network into the language of hardware, which takes time and energy. Building a network directly from hardware components does away with a lot of those costs. One day, they could even be built directly into chips used in smartphones and other devices, dramatically reducing the need to send data to and from servers. Felix Petersen, who did this work as a postdoctoral researcher at Stanford University, has a strategy for making that happen. He designed networks composed of logic gates, which are some of the basic building blocks of computer chips. Made up of a few transistors apiece, logic gates accept two bits1s or 0sas inputs and, according to a rule determined by their specific pattern of transistors, output a single bit. Just like perceptrons, logic gates can be chained up into networks. And running logic-gate networks is cheap, fast, and easy: in his talk at the Neural Information Processing Systems (NeurIPS) conference, Petersen said that they consume less energy than perceptron networks by a factor of hundreds of thousands. Logic-gate networks dont perform nearly as well as traditional neural networks on tasks like image labeling. But the approachs speed and efficiency make it promising, according to Zhiru Zhang, a professor of electrical and computer engineering at Cornell University. If we can close the gap, then this could potentially open up a lot of possibilities on this edge of machine learning, he says. Petersen didnt go looking for ways to build energy-efficient AI networks. He came to logic gates through an interest in differentiable relaxations, or strategies for wrangling certain classes of mathematical problems into a form that calculus can solve. It really started off as a mathematical and methodological curiosity, he says. Backpropagation, the training algorithm that made the deep-learning revolution possible, was an obvious use case for this approach. Because backpropagation runs on calculus, it cant be used directly to train logic-gate networks. Logic gates only work with 0s and 1s, and calculus demands answers about all the fractions in between. Petersen devised a way to relax logic-gate networks enough for backpropagation by creating functions that work like logic gates on 0s and 1s but also give answers for intermediate values. He ran simulated networks with those gates through training and then converted the relaxed logic-gate network back into something that he could implement in computer hardware. One challenge with this approach is that training the relaxed networks is tough. Each node in the network could end up as any one of 16 different logic gates, and the 16 probabilities associated with each of those gates must be kept track of and continually adjusted. That takes a huge amount of time and energyduring his NeurIPS talk, Petersen said that training his networks takes hundreds of times longer than training conventional neural networks on GPUs. At universities, which cant afford to amass hundreds of thousands of GPUs, that amount of GPU time can be tough to swingPetersen developed these networks, in collaboration with his colleagues, at Stanford University and the University of Konstanz. It definitely makes the research tremendously hard, he says. Once the network has been trained, though, things get way, way cheaper. Petersen compared his logic-gate networks with a cohort of other ultra-efficient networks, such as binary neural networks, which use simplified perceptrons that can process only binary values. The logic-gate networks did just as well as these other efficient methods at classifying images in the CIFAR-10 data set, which includes 10 different categories of low-resolution pictures, from frog to truck. It achieved this with fewer than a tenth of the logic gates required by those other methods, and in less than a thousandth of the time. Petersen tested his networks using programmable computer chips called FPGAs, which can be used to emulate many different potential patterns of logic gates; implementing the networks in non-programmable ASIC chips would reduce costs even further, because programmable chips need to use more components in order to achieve their flexibility. Farinaz Koushanfar, a professor of electrical and computer engineering at the University of California, San Diego, says she isnt convinced that logic-gate networks will be able to perform when faced with more realistic problems. Its a cute idea, but Im not sure how well it scales, she says. She notes that the logic-gate networks can only be trained approximately, via the relaxation strategy, and approximations can fail. That hasnt caused issues yet, but Koushanfar says that it could prove more problematic as the networks grow. Nevertheless, Petersen is ambitious. He plans to continue pushing the abilities of his logic-gate networks, and he hopes, eventually, to create what he calls a hardware foundation model. A powerful, general-purpose logic-gate network for vision could be mass-produced directly on computer chips, and those chips could be integrated into devices like personal phones and computers. That could reap enormous energy benefits, Petersen says. If those networks could effectively reconstruct photos and videos from low-resolution information, for example, then far less data would need to be sent between servers and personal devices. Petersen acknowledges that logic-gate networks will never compete with traditional neural networks on performance, but that isnt his goal. Making something that works, and that is as efficient as possible, should be enough. It wont be the best model, he says. But it should be the cheapest.

US doctors write billions of prescriptions each year. During 2024, though, one type of drug stood outwonder drugs known as GLP-1 agonists. As of September, one of every 20 prescriptions written for adults was for one of these drugs, according to the health data company Truveta. The drugs, which include Wegovy, Mounjaro, and Victoza, are used to treat diabetes, since they help generate insulin. But their popularity exploded after scientists determined the drugs tell your brain youre not hungry. Without those hunger cues, people find they can lose 10% of their body weight, or even more. During 2024, the drugs popularity hit an all-time high, according to Tricia Rodriguez, a principal applied scientist at Truveta, which studies medical records of 120 million Americans, or about a third of the population. Among adults, 5.4% of all prescriptions in September 2024 were for GLP-1s, Rodriguez says. That is up from 3.5% a year earlier, in 2023, and 1% at the start of 2021. According to Truvetas data, people who get prescriptions for these drugs are younger, whiter, and more likely to be female. In fact, women are twice as likely as men to get a prescription. Yet not everyone whos prescribed the drugs ends up taking them. In fact, Rodriguez says, half the new prescriptions for obesity are going unfilled. Thats very unusual, she says, and could be due to shortages or sticker shock over the cost of the treatment. Many insurers don t cover weight-loss drugs, and the out-of-pocket price can be $1,300 a month, according to USA Today. For most medications, prescribing rates and dispensing rates are pretty much identical, says Rodriguez. But for GLP-1s, we see this gap, which is really unique. It's suggestive that people are really interested in getting these medications, but for whatever reason, they are not always able to. It also means the number of people taking these drugs could go highermaybe much higherif insurers would pay. I don't think that we are at the saturation point, or necessarily nearing the saturation point, says Rodriguez, noting that around 70% of Americans are overweight or obese. Use of the drugs may also grow dramatically if new applications are found. Companies are already exploring whether they can treat addiction, or even Alzheimers. Many of the clues about those potential uses are coming directly out of peoples medical records. Because so many people are on the drugs, it means researchers like Rodriguez have a gold mine to sift through for signs of how use of the drugs is affecting other health problems. Because we have so many patients that are on these medications, you're certainly likely to have a good number that also have all of these other conditions, she says. One of the things we're excited about is: How can real-world data help accelerate how quickly we can understand those? Here are some of the new uses of GLP-1 drugs that are being explored, based on hints from real-world patient records. Alzheimers disease This year, researchers poking through records of a million people found that taking semaglutide (sold as Wegovy and Ozempic) was associated with a 40% to 70% lower chance of an Alzheimers diagnosis. It's still a guess why the drugs might be helping (or whether they really do), but large international studies are underway to follow up on the lead. Doctors are recruiting people with early Alzheimers in more than 30 countries who will take either a placebo or semaglutide for two years. Then well see how much their dementia has progressed. Addiction The anecdotes are everywhere: A person on a weight-loss drug finds hunger isnt the only craving that seems to stop. Those are the types of clues Eli Lillys CEO, David Ricks, says his company will pursue next year, testing whether its GLP-1 drug, tirzepatide (called Mounjaro for diabetes treatment, and Zepbound for weight loss), could help with addiction to alcohol, nicotine, and other things we dont think about [as being] connected to weight. In comments he made in December, Ricks said the drugs might be anti-hedonicsmeaning they counteract our hedonistic pursuit of pleasure, be it from food, alcohol, or drugs. A study this year mining digital health records found that opioid addicts taking the drugs were about half as likely to have had an overdose. Sleep apnea This idea goes back a ways, including to a 2015 case study of a 260-pound man with diabetes and sleep apnea. When he went on the drug liraglutide, doctors noticed that his sleeping improved. In sleep apnea, a person gasps for air at nightits annoying and, with time, causes health problems. This year, Eli Lilly published a study in the New England Journal of Medicine on its drugtirzepatide , finding that it caused a 50% decrease in breathing interruption in overweight patients with sleep apnea. Longevity This year, the U.S. Food and Drug Administration approved Wegovy as a cardiovascular medicine, after researchers showed the drugs could reduce heart attack and stroke in overweight people. But that wasnt all. The study, involving 17,000 people, found that the drug reduced the overall chance someone would die for any reason (known as all-cause mortality) by 19%. That now has aging researchers paying attention. This year they named Wegovy, and drugs like it, among their the top four candidates for a general life-extension drug.

A healthy heart beats at a steady rate, between 60 and 100 times a minute. Thats not the case for all of us, Im reminded, as I look inside a cardboard box containing around 20 plastic heartseach a replica of a real human one. The hearts, which previously sat on a shelf in a lab in West London, were generated from MRI and CT scans of people being treated for heart conditions at Hammersmith Hospital next door. Steven Niederer, a biomedical engineer at the Alan Turing Institute and Imperial College London, created them on a 3D printer in his office. One of the hearts, printed in red recycled plastic, looks as I imagine a heart to look. It just about fits in my hand, and the chambers have the same dimensions as the ones you might see in a textbook. Perhaps it helps that its red. The others look enormous to me. One in particular, printed in black plastic, seems more than twice the size of the red one. As I find out later, the person who had the heart it was modeled on suffered from heart failure. The plastic organs are just for educational purposes. Niederer is more interested in creating detailed replicas of peoples hearts using computers. These digital twins are the same size and shape as the real thing. They work in the same way. But they exist only virtually. Scientists can do virtual surgery on these virtual hearts, figuring out the best course of action for a patients condition. After decades of research, models like these are now entering clinical trials and starting to be used for patient care. Virtual replicas of many other organs are also being developed. Engineers are working on digital twins of peoples brains, guts, livers, nervous systems, and more. Theyre creating virtual replicas of peoples faces, which could be used to try out surgeries or analyze facial features, and testing drugs on digital cancers. The eventual goal is to create digital versions of our bodiescomputer copies that could help researchers and doctors figure out our risk of developing various diseases and determine which treatments might work best. Theyd be our own personal guinea pigs for testing out medicines before we subject our real bodies to them. To engineers like Niederer, its a tantalizing prospect very much within reach. Several pilot studies have been completed, and larger trials are underway. Those in the field expect digital twins based on organs to become a part of clinical care within the next five to 10 years, aiding diagnosis and surgical decision-making. Further down the line, well even be able to run clinical trials on synthetic patientsvirtual bodies created using real data. But the budding technology will need to be developed carefully. Some worry about who will own this highly personalized data and how it could be used. Others fear for patient autonomywith an uncomplicated virtual record to consult, will doctors eventually bypass the patients themselves? And some simply feel a visceral repulsion at the idea of attempts to re-create humans in silico. People will say I dont want you copying me, says Wahbi El-Bouri, who is working on digital-twin technologies. They feel its a part of them that youve taken. Getting digital Digital twins are well established in other realms of engineering; for example, they have long been used to model machinery and infrastructure. The term may have become a marketing buzzword lately, but for those working on health applications, it means something very specific. We can think of a digital twin as having three separate components, says El-Bouri, a biomedical engineer at the University of Liverpool in the UK. The first is the thing being modeled. That might be a jet engine or a bridge, or it could be a persons heart. Essentially, its what we want to test or study. The second component is the digital replica of that object, which can be created by taking lots of measurements from the real thing and entering them into a computer. For a heart, that might mean blood pressure recordings as well as MRI and CT scans. The third is new data thats fed into the model. A true digital twin should be updated in real timefor example, with information collected from wearable sensors, if its a model of someones heart. Taking measurements of airplanes and bridges is one thing. Its much harder to get a continuous data feed from a person, especially when you need details about the inner functions of the heart or brain. And the information transfer should run both ways. Just as sensors can deliver data from a persons heart, the computer can model potential outcomes to make predictions and feed them back to a patient or health-care provider. A medical team might want to predict how a person will respond to a drug, for example, or test various surgical procedures on a digital model before operating in real life. By this definition, pretty much any smart device that tracks some aspect of your health could be considered a kind of rudimentary digital twin. You could say that an Apple Watch fulfills the definition of a digital twin in an unexciting way, says Niederer. It tells you if youre in atrial fibrillation or not. But the kind of digital twin that researchers like Niederer are working on is far more intricate and detailed. It could provide specific guidance on which disease risks a person faces, what medicines might be most effective, or how any surgeries should proceed. Were not quite there yet. Taking measurements of airplanes and bridges is one thing. Its much harder to get a continuous data feed from a person, especially when you need details about the inner functions of the heart or brain, says Niederer. As things stand, engineers are technically creating patient-specific models based on previously collected hospital and research data, which is not continually updated. The most advanced medical digital twins are those built to match human hearts. These were the first to be attempted, partly because the heart is essentially a pumpa device familiar to engineersand partly because heart disease is responsible for so much ill health and death, says El-Bouri. Now, advances in imaging technology and computer processing power are enabling researchers to mimic the organ with the level of fidelity that clinical applications require. Building a heart The first step to building a digital heart is to collect images of the real thing. Each team will have its own slightly different approach, but generally, they all start with MRI and CT scans of a persons heart. These can be entered into computer software to create a 3D movie. Some scans will also highlight any areas of damaged tissue, which might disrupt the way the electrical pulses that control heart muscle contraction travel through the organ. The next step is to break this 3D model down into tiny chunks. Engineers use the term computational mesh to describe the result; it can look like an image of the heart made up of thousands of 3D pieces. Each segment represents a small collection of cells and can be assigned properties based on how well they are expected to propagate an electrical impulse. Its all equations, says Natalia Trayanova, a biomedical engineering professor based at Johns Hopkins University in Baltimore, Maryland. This computer model of the human heart show how electrical signals pass through heart tissue. The model was created by Marina Strocchi, who works with Steven Niederer at Imperial College London.COURTESY OF MARINA STROCCHI As things stand, these properties involve some approximation. Engineers will guess how well each bit of heart works by extrapolating from previous studies of human hearts or past research on the disease the person has. The end result is a beating, pumping model of a real heart. When we have that model, you can poke it and prod it and see under what circumstances stuff will happen, says Trayanova. Her digital twins are already being trialed to help people with atrial fibrillation, a fairly common condition that can trigger an irregular heartbeattoo fast or all over the place. One treatment option is to burn off the bits of heart tissue responsible for the disrupted rhythm. Its usually left to a surgical team to figure out which bits to target. For Trayanova, the pokes and prods are designed to help surgeons with that decision. Scans might highlight a few regions of damaged or scarred tissue. Her team can then construct a digital twin to help locate the underlying source of the damage. In total, the tool will likely suggest two or three regions to destroythough in rare instances, it has shown many more, says Trayanova: They just have to trust us. So far, 59 people have been through the trial. More are planned. In cases like these, the models dont always need to be continually updated, Trayanova says. A heart surgeon might need to run simulations only to know where to implant a device, for example. Once that operation is over, no more data might be needed, she says. Quasi patients At his lab on the campus of Hammersmith Hospital in London, Niederer has also been building virtual hearts. He is exploring whether his models could be used to find the best place to implant pacemakers. His approach is similar to Trayanovas, but his models also incorporate ECG data from patients. These recordings give a sense of how electrical pulses pass through the heart tissue, he says. So far, Niederer and his colleagues have published a small trial in which models of 10 patients hearts were evaluated by doctors but not used to inform surgical decisions. Still, Niederer is already getting requests from device manufacturers to run virtual tests of their products. A couple have asked him to choose places where their battery-operated pacemaker devices can sit without bumping into heart tissue, he says. Not only can Niederer and his colleagues run this test virtually, but they can do it for hearts of various different sizes. The team can test the device in hundreds of potential locations, within hundreds of different virtual hearts. And we can do it in a week, he adds. This is an example of what scientists call in silico trialsclinical trials run on a computer. In some cases, its not just the trials that are digital. The volunteers are, too. El-Bouri and his colleagues are working on ways to create synthetic participants for their clinical trials. The team starts with data collected from real people and uses this to create all-new digital organs with a mishmash of characteristics from the real volunteers. These in silico trials could be especially useful for helping us figure out the best treatments for pregnant peoplea group that is notoriously excluded from many clinical trials. Specifically, one of El-Bouris interests is stroke, a medical emergency in which clots or bleeds prevent blood flow in parts of the brain. For their research, he and his colleagues model the brain, along with the blood vessels that feed it. You could create lots and lots of different shapes and sizes of these brains based on patient data, says El-Bouri. Once he and his team create a group of synthetic patient brains, they can test how these clots might change the flow of blood or oxygen, or how and where brain tissue is affected. They can test the impact of certain drugs, or see what might happen if a stent is used to remove the blockage. For another project, El-Bouri is creating synthetic retinas. From a starting point of 100 or so retinal scans from real people, his team can generate 200 or more synthetic eyes, just like that, he says. The trick is to figure out the math behind the distribution of blood vessels and re-create it through a set of algorithms. Now he is hoping to use those synthetic eyes in drug trialsamong other things, to find the best treatment doses for people with age-related macular degeneration, a common condition that can lead to blindness. These in silico trials could be especially useful for helping us figure out the best treatments for pregnant peoplea group that is notoriously excluded from many clinical trials. Thats for fear that an experimental treatment might harm a fetus, says Michelle Oyen, a professor of biomedical engineering at Wayne State University in Detroit. Oyen is creating digital twins of pregnancy. Its a challenge to get the information needed to feed the models; during pregnancy, people are generally advised to avoid scans or invasive investigations they dont need. Were much more limited in terms of the data that we can get, she says. Her team does make use of ultrasound images, including a form of ultrasound that allows the team to measure blood flow. From those images, they can see how blood flow in the uterus and the placenta, the organ that supports a fetus, might be linked to the fetuss growth and development, for example. For now, Oyen and her colleagues arent creating models of the fetuses themselvestheyre focusing on the fetal environment, which includes the placenta and uterus. A baby needs a healthy, functioning placenta in order to survive; if the organ starts to fail, stillbirth can be the tragic outcome. Oyen is working on ways to monitor the placenta in real time during pregnancy. These readings could be fed back to a digital twin. If she can find a way to tell when the placenta is failing, doctors might be able to intervene to save the baby, she says. I think this is a game changer for pregnancy research, she adds, because this basically gives us ways of doing research in pregnancy that [carries a minimal] risk of harm to the fetus or of harm to the mother. In another project, the team is looking at the impact of cesarean section scars on pregnancies. When a baby is delivered by C-section, surgeons cut through multiple layers of tissue in the abdomen, including the uterus. Scars that dont heal well become weak spots in the uterus, potentially causing problems for future pregnancies. By modeling these scars in digital twins, Oyen hopes to be able to simulate how future pregnancies might pan out, and determine if or when specialist care might be called for. Eventually, Oyen wants to create a full virtual replica of the pregnant uterus, fetus and all. But were not there yetwere decades behind the cardiovascular people, she says. Thats pregnancy research in a nutshell, she adds. Were always decades behind. Twinning Its all very well to generate virtual body parts, but the human body functions as a whole. Thats why the grand plan for digital twins involves replicas of entire people. Long term, the whole body would be fantastic, says El-Bouri. It may not be all that far off, either. Various research teams are already building models of the heart, brain, lungs, kidneys, liver, musculoskeletal system, blood vessels, immune system, eye, ear, and more. If we were to take every research group that works on digital twins across the world at the moment, I think you could put [a body] together, says El-Bouri. I think theres even someone working on the tongue, he adds. The challenge is bringing together all the various researchers, with the different approaches and different code involved in creating and using their models, says El-Bouri. Everything exists, he says. Its just putting it together thats going to be the issue. In theory, such whole-body twins could revolutionize health care. Trayanova envisions a future in which a digital twin is just another part of a persons medical recordone that a doctor can use to decide on a course of treatment. Technically, if someone tried really hard, they might be able to piece back who someone is through scans and twins of organs. Wahbi El-Bouri But El-Bouri says he receives mixed reactions to the idea. Some people think its really exciting and really cool, he says. But hes also met people who are strongly opposed to the idea of having a virtual copy of themselves exist on a computer somewhere: They dont want any part of that. Researchers need to make more of an effort to engage with the public to find out how people feel about the technology, he says. There are also concerns over patient autonomy. If a doctor has access to a patients digital twin and can use it to guide decisions about medical care, where does the patients own input come into the equation? Some of those working to create digital twins point out that the models could reveal whether patients have taken their daily meds or what theyve eaten that week. Will clinicians eventually come to see digital twins as a more reliable source of information than peoples self-reporting? Doctors should not be allowed to bypass patients and just ask the machine, says Matthias Braun, a social ethicist at the University of Bonn in Germany. There would be no informed consent, which would infringe on autonomy and maybe cause harm, he says. After all, we are not machines with broken parts. Two individuals with the same diagnosis can have very different experiences and lead very different lives. However, there are cases in which patients are not able to make decisions about their own treatmentfor example, if they are unconscious. In those cases, clinicians try to find a proxysomeone authorized to make decisions on the patients behalf. A digital psychological twin, trained on a persons medical data and digital footprint, could potentially act as a better surrogate than, for example, a relative who doesnt know the persons preferences, he says. If using digital twins in patient care is problematic, in silico trials can also raise issues. Jantina de Vries, an ethicist at the University of Cape Town, points out that the data used to create digital twins and synthetic quasi patients will come from people who can be scanned, measured, and monitored. This group is unlikely to include many of those living on the African continent, who wont have ready access to those technologies. The problem of data scarcity directly translates into technologies that are not geared to think about diverse bodies, she says. De Vries thinks the data should belong to the public in order to ensure that as many people benefit from digital-twin technologies as possible. Every record should be anonymized and kept within a public database that researchers around the world can access and make use of, she says. The people who participate in Trayanovas trials explicitly give me consent to know their data, and to know who they are [everything] about them, she says. The people taking part in Niederers research also provide consent for their data to be used by the medical and research teams. But while clinicians have access to all medical data, researchers access only anonymized or pseudonymized data, Niederer says. In some cases, researchers will also ask participants to consent to sharing their fully anonymized data in public repositories. This is the only data that companies are able to access, he adds: We do not share [our] data sets outside of the research or medical teams, and we do not share them with companies. El-Bouri thinks that patients should receive some form of compensation in exchange for sharing their health data. Perhaps they should get preferential access to medications and devices based on that data, he suggests. At any rate, [full] anonymization is tricky, particularly if youre taking patient scans to develop twins, he says. Technically, if someone tried really hard, they might be able to piece back who someone is through scans and twins of organs. When I looked at those anonymous plastic hearts, stored in a cardboard box tucked away on a shelf in the corner of an office, they felt completely divorced from the people whose real, beating hearts they were modeled on. But digital twins seem different somehow. Theyre animated replicas, digital copies that certainly appear to have some sort of life. People often think, Oh, this is just a simulation, says El-Bouri. But its a digital representation of an individual.

This article is from The Spark, MIT Technology Reviews weekly climate newsletter. To receive it in your inbox every Wednesday, sign up here. The vibes in the climate world this year have largely been less than great. Global greenhouse-gas emissions hit a new high, reaching 37.4 billion metric tons in 2024. This year is also on track to be the warmest on record, with temperatures through September hitting 1.54 C (2.77 F) above preindustrial levels. Global climate talks fell flat, and disasters from wildfires to hurricanes are being made worse by climate change. But among all that (very real) negative news, there was some good, too: We saw progress cutting back on the most polluting fossil fuels, cheaper and better technologies for combating climate change, and a continuous global effort to address the problem. As we near the end of 2024, lets take a moment to look back on some of the bright spots. Were kicking coal to the curb One of my favorite climate moments from this year happened in the UK. The country has historically relied heavily on coal as an electricity sourceas of 1990, coal met about 65% of its electricity demand. But on September 30, 2024, the last coal plant in the nation shut down. Renewables are stepping in to fill the gap. Wind farms in the UK are on track to produce more electricity this year than coal and gas plants together. The moment was a symbolic one, and it also reflects the very real progress thats happening around the world in inching away from this polluting fossil fuel. In the US, coal made up around 50% of the electricity supply four decades ago. In 2023, that share was roughly 16%. We should see coal use plateau and potentially begin to fall by the end of the decade, according to the International Energy Agency. Progress needs to happen faster, though, and it needs to happen in countries like China, where energy demand is increasing. Theres also growing concern about what increasing energy demand from data centers, including those used to power AI, will mean for efforts to shut off old coal plants. Batteries just keep getting cheaper Lithium-ion battery packs are cheaper than ever in 2024, with prices dropping 20% this year to $115 per kilowatt-hour, according to data from BloombergNEF. Thats the biggest drop since 2017. Batteries are a central technology for addressing climate change. They power the electric vehicles were relying on to help clean up the transportation sector and play an increasingly important role for the grid, since they can store energy from inconsistently available renewables like wind and solar. Since EVs are still more expensive upfront than their gas-powered counterparts in most of the world, cheaper batteries are great news for efforts to get more people to take the leap to electric. And its hard to overstate how quickly battery prices have plummeted. Batteries were twice as expensive in 2017 as they are today. Just 10 years ago, prices were six times what they are in 2024. To be fair, theres been mixed news in the EV world this yeara slowdown in demand growth for EVs is actually one of the factors helping battery prices hit record lows. EV sales are still growing around the world, but at a slower pace than they were in 2023. China is the biggest EV market in the world by far, making up three-quarters of global registrations in 2024 as of October. Climate tech is still busy and bustling Looking back at the energy and climate stories we published this year, I cant help but feel at least a little bit optimistic about whats coming next. Some groups are looking to the natural world to address the climate crisis; this year, I covered a company working to grow microbes in massive bioreactors to help supplement our food sources, as well as researchers who are looking to plants to help mine the metals we need to fight climate change. Others hope to tweak biologymy colleague James Temple spoke with Jennifer Doudna about the potential for CRISPR, the gene-editing technology she pioneered. Companies are deploying air-conditioning systems that can act like batteries, storing up energy for when its needed. The US Department of Energy is investing in projects that aim to concentrate heat from the sun and use it to power the grid or industrial processes. I spoke to a startup looking to make hydropower technology thats safer for fish, and another building magnets using cheap, widely available materials. And in October we published our 2024 list of 15 Climate Tech Companies to Watch, which featured everything from a startup using AI to detect wildfires to a company giving supplements to cattle to help cut emissions from their burps. Climate change represents a massive challenge for the world, and were entering an especially uncertain time. Well be covering it all, the good and the bad. Thanks for being here this year, and Im looking forward to bringing you all the climate tech news you need in 2025. Now read the rest of The Spark Related reading If you need a dash of innovation and positivity in your life, might I recommend taking a gander at our list of 15 Climate Tech Companies to Watch? Whats more inspiring than young people working on the worlds most important problems? Our 2024 class of 35 Innovators Under 35 is sure to spread some cheer. If youre needing even more innovation, why not look back at our 10 Breakthrough Technologies? Exascale computers certainly help me put things in perspective. And get excited, because our 2025 list is coming very, very soon. NICO ORTEGA Another thing This year was filled with some exciting moments in technology, but there were also some failures. Here are a few of the worst technology flops of 2024. Check it out to see why voluntary carbon markets made the list and learn all about AI slop. And one more Youve almost certainly heard that energy demand from AI is huge, and only expected to explode in the coming years. A new preprint study aimed to quantify just how bad things are, and the researchers found that data centers accounted for over 4% of electricity consumption in the US between September 2023 and August 2024. And the carbon intensity of the power thats used is nearly 50% higher than the national average. Get all the details in the latest story from my colleague James ODonnell. Keeping up with climate Geothermal energy provides about 1% of global electricity today, but If things go well, the tech could meet up to 15% of global power demand growth through 2050. (Axios) Renting an EV over the holidays? This is a great guide for first-time EV drivers, including helpful tips about how to handle charging. (Bloomberg) Commonwealth Fusion Systems chose Virginia as the site for its first commercial fusion power plant. The company says the 400-megawatt plant will come online in the early 2030s. (Heatmap) I recently visited Commonwealths first demonstration site in Massachusetts. Its basically still a hole in the ground. (MIT Technology Review)The US Department of Energys Loan Programs Office just committed $15 billion to a California utility. Its the largest-ever commitment from the office. (New York Times) The US EPA will grant California the right to ban gas-powered cars by 2035. The agency has to give the state a waiver to set its own rules. (Washington Post) We can expect a legal battle, though. The incoming Trump administration is recommending major changes to cut off support for EVs and charging. (Reuters)China dominates the world of lithium-ion batteries. Some startups in the US and Europe argue that rather than playing catch-up, the rest of the world should focus on alternative chemistries like lithium-sulfur and sodium-ion batteries. (Canary Media)

This is today's edition ofThe Download,our weekday newsletter that provides a daily dose of what's going on in the world of technology. Digital twins of human organs are here. Theyre set to transform medical treatment. Steven Niederer, a biomedical engineer at the Alan Turing Institute and Imperial College London, has a cardboard box filled with 3D-printed hearts. Each of them is modeled on the real heart of a person with heart failure, but Niederer is more interested in creating detailed replicas of peoples hearts using computers. These digital twins are the same size and shape as the real thing. They work in the same way. But they exist only virtually. Scientists can do virtual surgery on these virtual hearts, figuring out the best course of action for a patients condition.After decades of research, models like these are now entering clinical trials and starting to be used for patient care. The eventual goal is to create digital versions of our bodiescomputer copies that could help researchers and doctors figure out our risk of developing various diseases and determine which treatments might work best.But the budding technology will need to be developed very carefully. Read the full story to learn why.Jessica Hamzelou This story is from the forthcoming magazine edition of MIT Technology Review, set to go live on January 6its all about the exciting breakthroughs happening in the world right now. If you dont already, subscribe to receive future copies. This is where the data to build AI comes from AI is all about data. Reams and reams of data are needed to train algorithms to do what we want, and what goes into the AI models determines what comes out. But heres the problem: AI developers and researchers dont really know much about the sources of the data they are using. The Data Provenance Initiative, a group of over 50 researchers from both academia and industry, wanted to fix that. They wanted to know, very simply: Where does the data to build AI come from? Their findings, shared exclusively with MIT Technology Review, show a worrying trend: AI's data practices risk concentrating power overwhelmingly in the hands of a few dominant technology companies. Read the full story.Melissa Heikkil Three pieces of good news on climate change in 2024 The vibes in the climate world this year have largely been less than great. Global greenhouse-gas emissions hit a new high, and this year is also on track to be the warmest on record. Global climate talks fell flat, and disasters from wildfires to hurricanes are being made worse by climate change. But among all that (very real) negative news, there was some good, too: We saw progress cutting back on the most polluting fossil fuels, cheaper and better technologies for combating climate change, and a continuous global effort to address the problem. So as we near the end of 2024, lets take a moment to look back on some of the bright spots. Casey Crownhart This story is from The Spark, our weekly climate and energy newsletter. Sign up to receive it in your inbox every Wednesday. The must-reads Ive combed the internet to find you todays most fun/important/scary/fascinating stories about technology. 1 The US Supreme Court will hear TikToks appeal against its ban Its agreed to hear the companys arguments on January 10. (FT $)+ A ruling could follow shortly afterwards. (WP $)+ Heres how a couple of the most likely scenarios could play out. (The Information $)2 Amazons telehealth clinic is being sued Philip Tong died shortly after a virtual appointment last year. His family wants answers. (WP $)+ The legal case accuses the health provider of negligently failing to care for Tong. (LA Times $)3 The Boeing Starliner astronauts are still stuck in space Their return to Earth has been pushed back yet again, this time to March 2025. (WP $)+ Theyve been living on the ISS since June. (The Guardian)4 Dangerous disordered eating content is rife on XThe platforms content moderation has become so lax, harmful communities are thriving unchecked. (The Atlantic $)5 People are shining lasers at planes flying over New York Amid the local drone panic, pilots are struggling with the unwelcome intrusions. (404 Media)+ Dont be surprised if other similar drone panics crop up in the future. (Vox)6 How Google Street View helped to solve a missing-person caseAfter its cars captured a man hunched over a large white bag in a car trunk. (NYT $) + Google Maps is still the biggest, but these startups are fast gaining traction. (Fast Company $)7 Why you shouldnt remove fluoride from your drinking waterUnless you desperately want to jeopardize your dental health. (WSJ $) + Its not the first time concerns around fluoride have surfaced. (NYT $) 8 The old internet is slowly disappearingWhat does that mean for our collective cultural understanding? (The Verge) + How to fix the internet. (MIT Technology Review)9 Europeans just love balcony solar panelsTheyre simple to install and can help to keep electricity bills down. (The Guardian) + How to store energy for leaner times. (Knowable Magazine)+ Advanced solar panels still need to pass the test of time. (MIT Technology Review)10 You can now call ChatGPT on the phone Theres nowhere left to hide. (Bloomberg $)Quote of the day I dont think that work is suitable for human beings. James Irungu, a former Facebook content moderator, reflects on the horrific material he encountered in the job, the Guardian reports. The big story Future space food could be made from astronaut breath May 2023 The future of space food could be as simpleand weirdas a protein shake made with astronaut breath or a burger made from fungus. For decades, astronauts have relied mostly on pre-packaged food during their forays off our planet. With missions beyond Earth orbit in sight, a NASA-led competition is hoping to change all that and usher in a new era of sustainable space food. To solve the problem of feeding astronauts on long-duration missions, NASA asked companies to propose novel ways to develop sustainable foods for future missions. Around 200 rose to the challengecreating nutritious (and outlandish) culinary creations in the process. Read the full story. Jonathan O'Callaghan We can still have nice things A place for comfort, fun and distraction to brighten up your day. (Got any ideas? Drop me a line or tweet 'em at me.) + These optical illusion bird sculptures are a sight to be seen. + Dont blame me if you end up wanting to eat this Bche de Nol in one sitting.+ Casio watches are 50 years oldand cooler than ever.+ Do you fly naked? (No, not like that..)

Business applications powered by AI are revolutionizing customer experiences, accelerating the speed of business, and driving employee productivity. In fact, according to research firm Frost & Sullivans 2024 Global State of AI report, 89% of organizations believe AI and machine learning will help them grow revenue, boost operational efficiency, and improve customer experience. Take for example, Vodafone. The telecommunications company is using a suite ofAzure AI services, such as Azure OpenAI Service,to deliver real-time, hyper-personalized experiences across all of its customer touchpoints, including its digital chatbot TOBi. By leveraging AI to increase customer satisfaction, Naga Surendran, senior director of product marketing for Azure Application Services at Microsoft, says Vodafone has managed to resolve 70% of its first-stage inquiries through AI-powered digital channels. It has also boosted the productivity of support agents by providing them with access to AI capabilities that mirror those of Microsoft Copilot, an AI-powered productivity tool. The result is a 20-point increase in net promotor score, he says. These benefits are whats driving AI infusion into every business process and application. Yet realizing measurable business value from AI-powered applications requires a new game plan. Legacy application architectures simply arent capable of meeting the high demands of AI-enhanced applications. Rather, the time is now for organizations to modernize their infrastructure, processes, and application architectures using cloud native technologies to stay competitive. The time is now for modernization Todays organizations exist in an era of geopolitical shifts, growing competition, supply chain disruptions, and evolving consumer preferences. AI applications can help by supporting innovation, but only if they have the flexibility to scale when needed. Fortunately, by modernizing applications, organizations can achieve the agile development, scalability, and fast compute performance needed to support rapid innovation and accelerate the delivery of AI applications. David Harmon, director of software development for AMD says companies, really want to make sure that they can migrate their current [environment] and take advantage of all the hardware changes as much as possible. The result is not only a reduction in the overall development lifecycle of new applications but a speedy response to changing world circumstances. Beyond building and deploying intelligent apps quickly, modernizing applications, data, and infrastructure can significantly improve customer experience. Consider, for example, Coles, an Australian supermarket thatinvested in modernization andis using data and AI to deliver dynamic e-commerce experiences to its customers both online and in-store. With Azure DevOps, Coles has shifted from monthly to weekly deployments of applications while, at the same time, reducing build times by hours. Whats more, by aggregating views of customers across multiple channels, Coles has been able to deliver more personalized customer experiences. In fact, according to a 2024 CMSWire Insights report, there is a significant rise in the use of AI across the digital customer experience toolset, with 55% of organizations now using it to some degree, and more beginning their journey. But even the most carefully designed applications are vulnerable to cybersecurity attacks. If given the opportunity, bad actors can extract sensitive information from machine learning models or maliciously infuse AI systems with corrupt data. AI applications are now interacting with your core organizational data, says Surendran. Having the right guard rails is important to make sure the data is secure and built on a platform that enables you to do that. The good news is modern cloud based architectures can deliver robust security, data governance, and AI guardrails like content safety to protect AI applications from security threats and ensure compliance with industry standards. The answer to AI innovation New challenges, from demanding customers to ill-intentioned hackers, call for a new approach to modernizing applications. You have to have the right underlying application architecture to be able to keep up with the market and bring applications faster to market, says Surendran. Not having that foundation can slow you down. Enter cloud native architecture. As organizations increasingly adopt AI to accelerate innovation and stay competitive, there is a growing urgency to rethink how applications are built and deployed in the cloud. By adopting cloud native architectures, Linux, and open source software, organizations can better facilitate AI adoption and create a flexible platform purpose built for AI and optimized for the cloud. Harmon explains that open source software creates options, And the overall open source ecosystem just thrives on that. It allows new technologies to come into play. Application modernization also ensures optimal performance, scale, and security for AI applications. Thats because modernization goes beyond just lifting and shifting application workloads to cloud virtual machines. Rather, a cloud native architecture is inherently designed to provide developers with the following features: The flexibility to scale to meet evolving needs Better access to the data needed to drive intelligent apps Access to the right tools and services to build and deploy intelligent applications easily Security embedded into an application to protect sensitive data Together, these cloud capabilities ensure organizations derive the greatest value from their AI applications. At the end of the day, everything is about performance and security, says Harmon. Cloud is no exception. Whats more, Surendran notes that when you leverage a cloud platform for modernization, organizations can gain access to AI models faster and get to market faster with building AI-powered applications. These are the factors driving the modernization journey. Best practices in play For all the benefits of application modernization, there are steps organizations must take to ensure both technological and operational success. They are: Train employees for speed. As modern infrastructure accelerates the development and deployment of AI-powered applications, developers must be prepared to work faster and smarter than ever. For this reason, Surendran warns, Employees must be skilled in modern application development practices to support the digital business needs. This includes developing expertise in working with loosely coupled microservices to build scalable and flexible application and AI integration. Start with an assessment. Large enterprises are likely to have hundreds of applications, if not thousands, says Surendran. As a result, organizations must take the time to evaluate their application landscape before embarking on a modernization journey. Starting with an assessment is super important, continues Surendran. Understanding, taking inventory of the different applications, which team is using what, and what this application is driving from a business process perspective is critical. Focus on quick wins. Modernization is a huge, long-term transformation in how companies build, deliver, and support applications. Most businesses are still learning and developing the right strategy to support innovation. For this reason, Surendran recommends focusing on quick wins while also working on a larger application estate transformation. You have to show a return on investment for your organization and business leaders, he says. For example, modernize some apps quickly with re-platforming and then infuse them with AI capabilities. Partner up. Modernization can be daunting, says Surendran. Selecting the right strategy, process, and platform to support innovation is only the first step. Organizations must also bring on the right set of partners to help them go through change management and the execution of this complex project. Address all layers of security. Organizations must be unrelenting when it comes to protecting their data. According to Surendran, this means adopting a multi-layer approach to security that includes: security by design, in which products and services are developed from the get-go with security in mind; security by default, in which protections exist at every layer and interaction where data exists; and security by ongoing operations, which means using the right tools and dashboards to govern applications throughout their lifecycle. A look to the future Most organizations are already aware of the need for application modernization. But with the arrival of AI comes the startling revelation that modernization efforts must be done right, and that AI applications must be built and deployed for greater business impact. Adopting a cloud native architecture can help by serving as a platform for enhanced performance, scalability, security, and ongoing innovation. As soon as you modernize your infrastructure with a cloud platform, you have access to these rapid innovations in AI models, says Surendran. Its about being able to continuously innovate with AI. Read more about how toaccelerate app and data estate readiness for AI innovationwithMicrosoft AzureandAMD. ExploreLinux on Azure. This content was produced by Insights, the custom content arm of MIT Technology Review. It was not written by MIT Technology Reviews editorial staff.

A small songbird soars above Ithaca, New York, on a September night. He is one of 4 billion birds, a great annual river of feathered migration across North America. Midair, he lets out what ornithologists call a nocturnal flight call to communicate with his flock. Its the briefest of signals, barely 50 milliseconds long, emitted in the woods in the middle of the night. But humans have caught it nevertheless, with a microphone topped by a focusing funnel. Moments later, software called BirdVoxDetect, the result of a collaboration between New York University, the Cornell Lab of Ornithology, and cole Centrale de Nantes, identifies the bird and classifies it to the species level. Biologists like Cornells Andrew Farnsworth had long dreamed of snooping on birds this way. In a warming world increasingly full of human infrastructure that can be deadly to them, like glass skyscrapers and power lines, migratory birds are facing many existential threats. Scientists rely on a combination of methods to track the timing and location of their migrations, but each has shortcomings. Doppler radar, with the weather filtered out, can detect the total biomass of birds in the air, but it cant break that total down by species. GPS tags on individual birds and careful observations by citizen-scientist birders help fill in that gap, but tagging birds at scale is an expensive and invasive proposition. And theres another key problem: Most birds migrate at night, when its more difficult to identify them visually and while most birders are in bed. For over a century, acoustic monitoring has hovered tantalizingly out of reach as a method that would solve ornithologists woes. In the late 1800s, scientists realized that migratory birds made species-specific nocturnal flight callsacoustic fingerprints. When microphones became commercially available in the 1950s, scientists began recording birds at night. Farnsworth led some of this acoustic ecology research in the 1990s. But even then it was challenging to spot the short calls, some of which are at the edge of the frequency range humans can hear. Scientists ended up with thousands of tapes they had to scour in real time while looking at spectrograms that visualize audio. Though digital technology made recording easier, the perpetual problem, Farnsworth says, was that it became increasingly easy to collect an enormous amount of audio data, but increasingly difficult to analyze even some of it. Then Farnsworth met Juan Pablo Bello, director of NYUs Music and Audio Research Lab. Fresh off a project using machine learning to identify sources of urban noise pollution in New York City, Bello agreed to take on the problem of nocturnal flight calls. He put together a team including the French machine-listening expert Vincent Lostanlen, and in 2015, the BirdVox project was born to automate the process. Everyone was like, Eventually, when this nut is cracked, this is going to be a super-rich source of information, Farnsworth says. But in the beginning, Lostanlen recalls, there was not even a hint that this was doable. It seemed unimaginable that machine learning could approach the listening abilities of experts like Farnsworth. Andrew is our hero, says Bello. The whole thing that we want to imitate with computers is Andrew. They started by training BirdVoxDetect, a neural network, to ignore faults like low buzzes caused by rainwater damage to microphones. Then they trained the system to detect flight calls, which differ between (and even within) species and can easily be confused with the chirp of a car alarm or a spring peeper. The challenge, Lostanlen says, was similar to the one a smart speaker faces when listening for its unique wake word, except in this case the distance from the target noise to the microphone is far greater (which means much more background noise to compensate for). And, of course, the scientists couldnt choose a unique sound like Alexa or Hey Google for their trigger. For birds, we dont really make that choice. Charles Darwin made that choice for us, he jokes. Luckily, they had a lot of training data to work withFarnsworths team had hand-annotated thousands of hours of recordings collected by the microphones in Ithaca. With BirdVoxDetect trained to detect flight calls, another difficult task lay ahead: teaching it to classify the detected calls by species, which few expert birders can do by ear. To deal with uncertainty, and because there is not training data for every species, they decided on a hierarchical system. For example, for a given call, BirdVoxDetect might be able to identify the birds order and family, even if its not sure about the speciesjust as a birder might at least identify a call as that of a warbler, whether yellow-rumped or chestnut-sided. In training, the neural network was penalized less when it mixed up birds that were closer on the taxonomical tree. Last August, capping off eight years of research, the team published a paper detailing BirdVoxDetects machine-learning algorithms. They also released the software as a free, open-source product for ornithologists to use and adapt. In a test on a full season of migration recordings totaling 6,671 hours, the neural network detected 233,124 flight calls. In a 2022 study in the Journal of Applied Ecology, the team that tested BirdVoxDetect found acoustic data as effective as radar for estimating total biomass. BirdVoxDetect works on a subset of North American migratory songbirds. But through few-shot learning, it can be trained to detect other, similar birds with just a few training examples. Its like learning a language similar to one you already speak, Bello says. With cheap microphones, the system could be expanded to places around the world without birders or Doppler radar, even in vastly different recording conditions. If you go to a bioacoustics conference and you talk to a number of people, they all have different use cases, says Lostanlen. The next step for bioacoustics, he says, is to create a foundation model, like the ones scientists are working on for natural-language processing and image and video analysis, that would be reconfigurable for any specieseven beyond birds. That way, scientists wont have to build a new BirdVoxDetect for every animal they want to study. The BirdVox project is now complete, but scientists are already building on its algorithms and approach. Benjamin Van Doren, a migration biologist at the University of Illinois Urbana-Champaign who worked on BirdVox, is using Nighthawk, a new user-friendly neural network based on both BirdVoxDetect and the popular birdsong ID app Merlin, to study birds migrating over Chicago and elsewhere in North and South America. And Dan Mennill, who runs a bioacoustics lab at the University of Windsor, says hes excited to try Nighthawk on flight calls his team currently hand-annotates after theyre recorded by microphones on the Canadian side of the Great Lakes. One weakness of acoustic monitoring is that unlike radar, a single microphone cant detect the altitude of a bird overhead or the direction in which it is moving. Mennills lab is experimenting with an array of eight microphones that can triangulate to solve that problem. Sifting through recordings has been slow. But with Nighthawk, the analysis will speed dramatically. With birds and other migratory animals under threat, Mennill says, BirdVoxDetect came at just the right time. Knowing exactly which birds are flying over in real time can help scientists keep tabs on how species are doing and where theyre going. That can inform practical conservation efforts like Lights Out initiatives that encourage skyscrapers to go dark at night to prevent bird collisions. Bioacoustics is the future of migration research, and were really just getting to the stage where we have the right tools, he says. This ushers us into a new era. Christian Elliott is a science and environmental reporter based in Illinois.

AI is all about data. Reams and reams of data are needed to train algorithms to do what we want, and what goes into the AI models determines what comes out. But heres the problem: AI developers and researchers dont really know much about the sources of the data they are using. AIs data collection practices are immature compared with the sophistication of AI model development. Massive data sets often lack clear information about what is in them and where it came from. The Data Provenance Initiative, a group of over 50 researchers from both academia and industry, wanted to fix that. They wanted to know, very simply: Where does the data to build AI come from? They audited nearly 4,000 public data sets spanning over 600 languages, 67 countries, and three decades. The data came from 800 unique sources and nearly 700 organizations. Their findings, shared exclusively with MIT Technology Review, show a worrying trend: AI's data practices risk concentrating power overwhelmingly in the hands of a few dominant technology companies. In the early 2010s, data sets came from a variety of sources, says Shayne Longpre, a researcher at MIT who is part of the project. It came not just from encyclopedias and the web, but also from sources such as parliamentary transcripts, earning calls, and weather reports. Back then, AI data sets were specifically curated and collected from different sources to suit individual tasks, Longpre says. Then transformers, the architecture underpinning language models, were invented in 2017, and the AI sector started seeing performance get better the bigger the models and data sets were. Today, most AI data sets are built by indiscriminately hoovering material from the internet. Since 2018, the web has been the dominant source for data sets used in all media, such as audio, images, and video, and a gap between scraped data and more curated data sets has emerged and widened. In foundation model development, nothing seems to matter more for the capabilities than the scale and heterogeneity of the data and the web, says Longpre. The need for scale has also boosted the use of synthetic data massively. The past few years have also seen the rise of multimodal generative AI models, which can generate videos and images. Like large language models, they need as much data as possible, and the best source for that has become YouTube. For video models, as you can see in this chart, over 70% of data for both speech and image data sets comes from one source. This could be a boon for Alphabet, Googles parent company, which owns YouTube. Whereas text is distributed across the web and controlled by many different websites and platforms, video data is extremely concentrated in one platform. It gives a huge concentration of power over a lot of the most important data on the web to one company, says Longpre. And because Google is also developing its own AI models, its massive advantage also raises questions about how the company will make this data available for competitors, says Sarah Myers West, the coexecutive director at the AI Now Institute. Its important to think about data not as though its sort of this naturally occurring resource, but its something that is created through particular processes, says Myers West. If the data sets on which most of the AI that were interacting with reflect the intentions and the design of big, profit-motivated corporationsthats reshaping the infrastructures of our world in ways that reflect the interests of those big corporations, she says. This monoculture also raises questions about how accurately the human experience is portrayed in the data set and what kinds of models we are building, says Sara Hooker, the vice president of research at the technology company Cohere, who is also part of the Data Provenance Initiative. People upload videos to YouTube with a particular audience in mind, and the way people act in those videos is often intended for very specific effect. Does [the data] capture all the nuances of humanity and all the ways that we exist? says Hooker. Hidden restrictions AI companies dont usually share what data they used to train their models. One reason is that they want to protect their competitive edge. The other is that because of the complicated and opaque way data sets are bundled, packaged, and distributed, they likely dont even know where all the data came from. They also probably dont have complete information about any constraints on how that data is supposed to be used or shared. The researchers at the Data Provenance Initiative found that data sets often have restrictive licenses or terms attached to them, which should limit their use for commercial purposes, for example. This lack of consistency across the data lineage makes it very hard for developers to make the right choice about what data to use, says Hooker. It also makes it almost impossible to be completely certain you havent trained your model on copyrighted data, adds Longpre. More recently, companies such as OpenAI and Google have struck exclusive data-sharing deals with publishers, major forums such as Reddit, and social media platforms on the web. But this becomes another way for them to concentrate their power. These exclusive contracts can partition the internet into various zones of who can get access to it and who cant, says Longpre. The trend benefits the biggest AI players, who can afford such deals, at the expense of researchers, nonprofits, and smaller companies, who will struggle to get access. The largest companies also have the best resources for crawling data sets. This is a new wave of asymmetric access that we havent seen to this extent on the open web, Longpre says. The West vs. the rest The data that is used to train AI models is also heavily skewed to the Western world. Over 90% of the data sets that the researchers analyzed came from Europe and North America, and fewer than 4% came from Africa. "These data sets are reflecting one part of our world and our culture, but completely omitting others," says Hooker. The dominance of the English language in training data is partly explained by the fact that the internet is still over 90% in English, and there are still a lot of places on Earth where theres really poor internet connection or none at all, says Giada Pistilli, principal ethicist at Hugging Face, who was not part of the research team. But another reason is convenience, she adds: Putting together data sets in other languages and taking other cultures into account requires conscious intention and a lot of work. The Western focus of these data sets becomes particularly clear with multimodal models. When an AI model is prompted for the sights and sounds of a wedding, for example, it might only be able to represent Western weddings, because thats all that it has been trained on, Hooker says. This reinforces biases and could lead to AI models that push a certain US-centric worldview, erasing other languages and cultures. We are using these models all over the world, and theres a massive discrepancy between the world were seeing and whats invisible to these models, Hooker says.

This is today's edition ofThe Download,our weekday newsletter that provides a daily dose of what's going on in the world of technology. AI is changing how we study bird migration In a warming world, migratory birds face many existential threats. Scientists rely on a combination of methods to track the timing and location of their migrations, but each has shortcomings. And theres another problem: Most birds migrate at night, when its more difficult to identify them visually and while most birders are in bed. For over a century, acoustic monitoring has hovered tantalizingly out of reach as a method that would solve ornithologists woes. Now, finally, machine-learning tools are unlocking a treasure trove of acoustic data for ecologists. Read the full story.Christian Elliot This story is from the forthcoming magazine edition of MIT Technology Review, set to go live on January 6its all about the exciting breakthroughs happening in the world right now. If you dont already, subscribe to receive a copy. A woman in the US is the third person to receive a gene-edited pig kidney Towana Looney, a 53-year-old woman from Alabama, has become the third living person to receive a kidney transplant from a gene-edited pig. Looney, who donated one of her kidneys to her mother back in 1999, developed kidney failure several years later following a pregnancy complication that caused high blood pressure. She started dialysis treatment in December of 2016 and was put on a waiting list for a kidney transplant soon after. But it was difficult to find a match. So Looneys doctors recommended the experimental pig organ as an alternative. After eight years on the waiting list, Looney was authorized to receive the kidney. Read the full story. Jessica Hamzelou Roundtables: The Worst Technology Failures of 2024 Each year, MIT Technology Review publishes a list of the worst technologies of the past 12 months. Antonio Regalado, our senior editor for biomedicine, sat down to discuss 2024s worst failures with our executive editor Niall Firth in a subscriber-exclusive online Roundtable event yesterday. Watch their conversation about what made the cut here, and to make sure you dont miss out in the future, subscribe!MIT Technology Review Narrated: Meet the radio-obsessed civilian shaping Ukraines drone defense Despite it being over 100 years old, radio technology is still critical in almost all aspects of modern warfareincluding in the drones that have come to dominate the Russia-Ukraine war. Serhii Flash Beskrestnov, who has been obsessed with radios since childhood, has become an unlikely hero of the conflict, sharing advice and intel. His work may determine the future of Ukraine, and wars far beyond it. This is our latest story to be turned into a MIT Technology Review Narrated podcast, which were publishing each week on Spotify and Apple Podcasts. Just navigate to MIT Technology Review Narrated on either platform, and follow us to get all our new content as its released.The must-reads Ive combed the internet to find you todays most fun/important/scary/fascinating stories about technology. 1 Conspiracy theories are still circulating about those mysterious drones What are they? And where have they come from? (NY Mag $)+ Authorities are attempting to quell public hysteria, but theories abound. (WP $)+ Realistically, theyre probably just standard drones out for a night-time flight. (AP News)2 AI poses a major threat to the power grid Thats according to the US industry watchdog, which is feeling the pressure. (FT $)+ AIs emissions are about to skyrocket even further. (MIT Technology Review)3 SpaceX and Elon Musk are under investigation US federal agencies are probing their repeated failures to comply with reporting rules. (NYT $)4 Nvidia has unveiled a tiny, affordable AI supercomputer Which is handy for roboticists looking to bypass connecting to remote data centers. (Gizmodo)+ While its not the companys most powerful device, its pretty speedy. (WSJ $)+ Microsoft is gobbling up more of Nvidias chips than anyone else. (FT $)+ Blacklisted Chinese AI chip firms gained access to cutting-edge UK tech. (The Guardian) 5 Bitcoins value is rocketing even higherThe industry continues to boom in the wake of Trumps election victory. (Bloomberg $) + So much so, luxury brands are weighing up accepting crypto payments. (Reuters)6 Hepatitis B is an extremely treatable diseaseSo why are so many people still dying from it? (New Yorker $) + Were starting to understand the mysterious surge of hepatitis in children. (MIT Technology Review)7 Earthbrieflyhad an extra second moon And scientists believe it originated from the actual moon we know and love. (New Scientist $) 8 The future of deep-sea miningA set of rules governing how we should do it is highly contentiousand up for debate.(Hakai Magazine) + These deep-sea potatoes could be the future of mining for renewable energy. (MIT Technology Review)9 Resist the temptation to outsource your Christmas shopping to a bot You never know what youll end up with. (Insider $)+ Its probably quicker to browse the web yourself. (WP $)10 Our snacks could soon be designed by AI Confectionary giant Mondelez is using the tech to tweak recipes and test new ones. (WSJ $)+ Forget cookiesthis creamy vegan cheese was made with AI. (MIT Technology Review) Quote of the day It takes a lot for an uber-wealthy, creative-type CEO, many of whom lean left, to suck it up and deal with Trump. But what choice do they have? A Washington lobbyist explains to the Financial Times why the steady stream of tech executives paying their respects to US President-elect Donald Trump shows no sign of slowing. The big story What does GPT-3 know about me? August 2022 One of the biggest stories in tech is the rise of large language models that produce text that reads like a human might have written it. These models power comes from being trained on troves of publicly available human-created text hoovered up from the internet. If youve posted anything even remotely personal in English on the internet, chances are your data might be part of some of the worlds most popular LLMs. Melissa Heikkil, MIT Technology Reviews AI reporter, wondered what data these models might have on herand how it could be misused. So she put OpenAIs GPT-3 to the test. Read about what she found.We can still have nice things A place for comfort, fun and distraction to brighten up your day. (Got any ideas? Drop me a line or tweet 'em at me.) + 2024 was a seriously weird year, as evidenced by this completely bonkers list.+ Who knew Seal was such a grunge head?+ These Charli xcx Christmas mashups will haunt my dreams forever, and not in a good way.+ Next summer I feel the need to level up my sandcastle game.

This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here. If you drove by one of the 2,990 data centers in the United States, youd probably think little more than Huh, thats a boring-looking building. You might not even notice it at all. However, these facilities underpin our entire digital world, and they are responsible for tons of greenhouse-gas emissions. New research shows just how much those emissions have skyrocketed during the AI boom. Since 2018, carbon emissions from data centers in the US have tripled, according to new research led by a team at the Harvard T.H. Chan School of Public Health. That puts data centers slightly below domestic commercial airlines as a source of this pollution. That leaves a big problem for the worlds leading AI companies, which are caught between pressure to meet their own sustainability goals and the relentless competition in AI thats leading them to build bigger models requiring tons of energy. The trend toward ever more energy-intensive new AI models, including video generators like OpenAIs Sora, will only send those numbers higher. A growing coalition of companies is looking toward nuclear energy as a way to power artificial intelligence. Meta announced on December 3 it was looking for nuclear partners, and Microsoft is working to restart the Three Mile Island nuclear plant by 2028. Amazon signed nuclear agreements in October. However, nuclear plants take ages to come online. And though public support has increased in recent years, and president-elect Donald Trump has signaled support, only a slight majority of Americans say they favor more nuclear plants to generate electricity. Though OpenAI CEO Sam Altman pitched the White House in September on an unprecedented effort to build more data centers, the AI industry is looking far beyond the United States. Countries in Southeast Asia, like Malaysia, Indonesia, Thailand, and Vietnam, are all courting AI companies, hoping to be their new data center hubs. In the meantime, AI companies will continue to use up power from their current sources, which are far from renewable. Since so many data centers are located in coal-producing regions, like Virginia, the carbon intensity of the energy they use is 48% higher than the national average. The researchers found that 95% of data centers in the US are built in places with sources of electricity that are dirtier than the national average. Read more about the new research here. Deeper Learning We saw a demo of the new AI system powering Andurils vision for war Were living through the first drone wars, but AI is poised to change the future of warfare even more drastically. I saw that firsthand during a visit to a test site in Southern California run by Anduril, the maker of AI-powered drones, autonomous submarines, and missiles. Anduril has built a way for the military to command much of its hardwarefrom drones to radars to unmanned fighter jetsfrom a single computer screen. Why it matters: Anduril, other companies in defense tech, and growing numbers of people within the Pentagon itself are increasingly adopting a new worldview: A future great power conflictmilitary jargon for a global war involving multiple countrieswill not be won by the entity with the most advanced drones or firepower, or even the cheapest firepower. It will be won by whoever can sort through and share information the fastest. The Pentagon is betting lots of energy and money that AIdespite its flaws and riskswill be what puts the US and its allies ahead in that fight. Read more here. Bits and Bytes Bluesky has an impersonator problem The platforms rise has brought with it a surge of crypto scammers, as my colleague Melissa Heikkil experienced firsthand. (MIT Technology Review) Techs elite make large donations to Trump ahead of his inauguration Leaders in Big Tech, who have been lambasted by Donald Trump, have made sizable donations to his inauguration committee. (The Washington Post) Inside the premiere of the first commercially streaming AI-generated movies The films, according to writer Jason Koebler, showed the telltale flaws of AI-generated video: dead eyes, vacant expressions, unnatural movements, and a reliance on voice-overs, since dialogue doesnt work well. The company behind the films is confident viewers will stomach them anyway. (404 Media) Meta asked Californias attorney general to stop OpenAI from becoming for-profit Meta now joins Elon Musk in alleging that OpenAI has improperly enjoyed the benefits of nonprofit status while developing its technology. (Wall Street Journal) How Silicon Valley is disrupting democracy Two books explore the price weve paid for handing over unprecedented power to Big Techand explain why its imperative we start taking it back. (MIT Technology Review)

They say you learn more from failure than success. If so, this is the story for you: MIT Technology Reviews annual roll call of the biggest flops, flimflams, and fiascos in all domains of technology. Some of the foul-ups were funny, like the "woke AI which got Google in trouble after it drew Black Nazis. Some caused lawsuits, like a computer error by CrowdStrike that left thousands of Delta passengers stranded. We also reaped failures among startups that raced to expand from 2020 to 2022, a period of ultra-low interest rates. But then the economic winds shifted. Money wasnt free anymore. The result? Bankruptcy and dissolution for companies whose ambitious technological projects, from vertical farms to carbon credits, hadnt yet turned a profit and might never do so. Read on. Woke AI blunder GOOGLE GEMINI VIA X.COM/END WOKENESS People worry about bias creeping into AI. But what if you add bias on purpose? Thanks to Google, we know where that leads: Black Vikings and female popes. Googles Gemini AI image feature, launched last February, had been tuned to zealously showcase diversity, damn the history books. Ask Google for a picture of German soldiers from World War II, and it would create a Benetton ad in Wehrmacht uniforms. Critics pounced and Google beat an embarrassed retreat. It paused Geminis ability to draw people and agreed its well-intentioned effort to be inclusive had missed the mark. The free version of Gemini still wont create images of people. But paid versions will. When we asked for an image of 12 CEOs of public biotech companies, the software produced a photographic-quality image of middle-aged white men. Less than ideal. But closer to the truth. More: Is Googles Gemini chatbot woke by accident, or by design? (The Economist), Gemini image generation got it wrong. We'll do better. (Google) Boeing Starliner THE BOEING COMPANY VIA NASA Boeing, we have a problem. And its your long-delayed reusable spaceship, the Starliner, which stranded NASA astronauts Sunita Suni Williams and Barry Butch Wilmore on the International Space Station. The June mission was meant to be a quick eight-day round trip to test Starliner before it embarked on longer missions. But, plagued by helium leaks and thruster problems, it had to come back empty. Now Butch and Suni wont return to Earth until 2025, when a craft from Boeing competitor SpaceX is scheduled to bring them home. Credit Boeing and NASA with putting safety first. But this wasnt Boeings only malfunction during 2024. The company began the year with a door blowing off one of its planes midflight, faced a worker strike, agreed to a major fine for misleading the government about the safety of its 737 Max airplane (which made our 2019 list of worst technologies), and saw its CEO step down in March. After the Starliner fiasco, Boeing fired the chief of its space and defense unit. At this critical juncture, our priority is to restore the trust of our customers and meet the high standards they expect of us to enable their critical missions around the world, Boeings new CEO, Kelly Ortberg, said in a memo. CrowdStrike outage MITTR / ENVATO The motto of the cybersecurity company CrowdStrike is We stop breaches. And its true: No one can breach your computer if you cant turn it on. Thats exactly what happened to many people on July 19, when thousands of Windows computers at airlines, TV stations, and hospitals started displaying the blue screen of death. The cause wasnt hackers or ransomware. Instead, those computers were stuck in a boot loop because of a bad update shipped by CrowdStrike itself. CEO George Kurtz jumped on X to say the issue had been identified as a defect in a single computer file. So who is liable? CrowdStrike customer Delta Airlines, which canceled 7,000 flights, is suing for $500 million. It alleges that the security firm caused a global catastrophe when it took uncertified and untested shortcuts. CrowdStrike countersued. It says Deltas management is to blame for its troubles and that the airline is due little more than a refund. More: Crowdstrike is working with customers(George Kurtz), How to fix a Windows PC affected by the global outage (MIT Technology Review), Delta Sues CrowdStrike Over July Operations Meltdown (WSJ) Vertical farms MITTR / ENVATO Grow lettuce in buildings using robots, hydroponics, and LED lights. Thats what Bowery, a vertical farming startup, raised over $700 million to do. But in November, Bowery went bust, making it the biggest startup failure of the year, according to the business analytics firm CB Insights. Bowery claimed that vertical farms were 100 times more productive per square foot than traditional farms, since racks of plants could be stacked 40 feet high. In reality, the companys lettuce was more expensive, and when a stubborn plant infection spread through its East Coast facilities, Bowery had trouble delivering the green stuff at any price. More: How a leaf-eating pathogen, failed deals brought down Bowery Farming (Pitchbook), Vertical farming "unicorn" Bowery to shut down (Axios) Exploding pagers MITTR / ADOBE STOCK They beeped, and then they blew up. Across Lebanon, fingers and faces were shredded in what was called Israels surprise opening blow in an all-out war to try to cripple Hezbollah. The deadly attack was diabolically clever. Israel set up shell companies that sold thousands of pagers packed with explosives to the Islamic faction, which was already worried that its phones were being spied on. A coup for Israels spies. But was it a war crime? A 1996 treaty prohibits intentionally manufacturing apparently harmless objects designed to explode. The New York Times says nine-year-old Fatima Abdullah died when her fathers booby-trapped beeper chimed and she raced to take it to him. More: Israel conducted Lebanon pager attack (Axios), A 9-Year-Old Girl Killed in Pager Attack Is Mourned in Lebanon (New York Times), Did Israel break international law? (Middle East Eye) 23andMe MITTR / ADOBE STOCK The company that pioneered direct-to-consumer gene testing is sinking fast. Its stock price is going toward zero, and a plan to create valuable drugs is kaput after that team got pink slips this November. 23andMe always had a celebrity aura, bathing in good press. Now, though, the press is all bad. Its a troubled company in the grip of a controlling founder, Anne Wojcicki, after its independent directors resigned en masse this September. Customers are starting to worry about whats going to happen to their DNA data if 23andMe goes under. 23andMe says it created the worlds largest crowdsourced platform for genetic research. Thats true. It just never figured out how to turn a profit. More: 23andMes fall from $6 billion to nearly $0 (Wall Street Journal), How todelete your 23andMe data (MIT Technology Review), 23andMe Financial Report, November 2024 (23andMe) AI slop AUTHOR UNKNOWN VIA WIKIMEDIA COMMONS Slop is the scraps and leftovers that pigs eat. AI slop is what you and I are increasingly consuming online now that people are flooding the internet with computer-generated text and pictures. AI slop is dubious, says the New York Times, and dadaist, according to Wired. Its frequently weird, like Shrimp Jesus (dont ask if you dont know), or deceptive, like the picture of a shivering girl in a rowboat, supposedly showing the US governments poor response to Hurricane Helene. AI slop is often entertaining. AI slop is usually a waste of your time. AI slop is not fact-checked. AI slop exists mostly to get clicks. AI slop is that blue-check account on X posting 10-part threads on how great AI isthreads that were written by AI. Most of all, AI slop is very, very common. This year, researchers claimed that about half the long posts on LinkedIn and Medium were partly AI-generated. More: First came Spam. Now, With A.I., Weve got Slop (New York Times), AI Slop Is Flooding Medium (Wired) Voluntary carbon markets MITTR / ENVATO Your business creates emissions that contribute to global warming. So why not pay to have some trees planted or buy a more efficient cookstove for someone in Central America? Then you could reach net-zero emissions and help save the planet. Neat idea, but good intentions aren't enough. This year the carbon marketplace Nori shut down, and so did Running Tide, a firm trying to sink carbon into the ocean. The problem is the voluntary carbon market is voluntary, Running Tides CEO wrote in a farewell post, citing a lack of demand. While companies like to blame low demand, it's not the only issue. Sketchy technology, questionable credits, and make-believe offsets have created a credibility problem in carbon markets. In October, US prosecutors charged two men in a $100 million scheme involving the sale of nonexistent emissions savings. More: The growing signs of trouble for global carbon markets (MIT Technology Review), Running Tides ill-fated adventure in ocean carbon removal (Canary Media), Ex-carbon offsetting boss charged in New York with multimillion-dollar fraud (The Guardian)

Recorded on December 17, 2024The Worst Technology Failures of 2024Speakers: Antonio Regalado, senior editor for biomedicine, and Niall Firth, executive editor.MIT Technology Review publishes an annual list of the worst technologies of the year. This year, The Worst Technology Failures of 2024 list was unveiled live by our editors. Hear fromMIT Technology Reviewexecutive editor Niall Firth and senior editor for biomedicine Antonio Regalado as they discuss each of the 8 items on this list.Related CoverageThe 8 worst technology failures of 2024The worst technology failures of 2023The worst technology of 2022

This is today's edition ofThe Download,our weekday newsletter that provides a daily dose of what's going on in the world of technology. The 8 worst technology failures of 2024 They say you learn more from failure than success. If so, this is the story for you: MIT Technology Reviews annual roll call of the biggest flops, flimflams, and fiascos in all domains of technology. Some of the foul-ups were funny, like the "woke AI which got Google in trouble after it drew Black Nazis. Some caused lawsuits, like a computer error by CrowdStrike that left thousands of Delta passengers stranded. And we also reaped failures among startups that raced to expand from 2020 to 2022, a period of ultra-low interest rates. Check out what made our list of this years biggest technology failures. Antonio Regalado Antonio will be discussing this years worst failures with our executive editor Niall Firth in a subscriber-exclusive online Roundtable event today at 12.00 ET. Register here to make sure you dont miss outf you havent already, subscribe! AIs search for more energy is growing more urgent If you drove by one of the 2,990 data centers in the United States, youd probably think little more than Huh, thats a boring-looking building. You might not even notice it at all. However, these facilities underpin our entire digital world, and they are responsible for tons of greenhouse-gas emissions. New research shows just how much those emissions have skyrocketed during the AI boom. That leaves a big problem for the worlds leading AI companies, which are caught between pressure to meet their own sustainability goals and the relentless competition in AI thats leading them to build bigger models requiring tons of energy. And the trend toward ever more energy-intensive new AI models will only send those numbers higher. Read the full story.James O'Donnell This story originally appeared in The Algorithm, our weekly newsletter on AI. Sign up to receive it in your inbox every Monday. The must-reads Ive combed the internet to find you todays most fun/important/scary/fascinating stories about technology. 1 TikTok has asked the US Supreme Court for a lifeline Its asked lawmakers to intervene before the proposed ban kicks in on January 19. (WP $)+ TikTok CEO Shou Zi Chew reportedly met with Donald Trump yesterday. (NBC News)+ Trump will take office the following day, on January 20. (WSJ $)+ Meanwhile, the EU is investigating TikToks role in Romanias election. (Politico)2 Waymos autonomous cars are heading to Tokyo In the first overseas venture for the firms vehicles. (The Verge)+ The cars will require human safety drivers initially. (CNBC)+ Whats next for robotaxis in 2024. (MIT Technology Review)3 Chinas tech workers are still keen to work in the US But securing the right to work there is much tougher than it used to be. (Rest of World)4 Digital license plates are vulnerable to hacking And theyre already legal to buy in multiple US states. (Wired $) 5 Were all slaves to the algorithmsFrom the mundane (Spotify) to the essential (housing applications.) (The Atlantic $) + How a group of tenants took on screening systemsand won. (The Guardian)+ The coming war on the hidden algorithms that trap people in poverty. (MIT Technology Review)6 How to build an undetectable submarine The race is on to stay hidden from the competition. (IEEE Spectrum)+ How underwater drones could shape a potential Taiwan-China conflict. (MIT Technology Review)7 How Empower became a viable rival to UberIts refusal to cooperate with authorities is straight out of Ubers early playbook. (NYT $) 8 Even airlines are using AirTags to find lost luggage Bloomberg $) + Heres how to keep tabs on your suitcase as you travel. (Forbes $)9 Youre reading your blood pressure all wrong Keep your feet flat on the floor and ditch your phone, for a start. (WSJ $)10 The rise and rise of the group chat Expressing yourself publicly on social media is so last year. (Insider $)+ How to fix the internet. (MIT Technology Review)Quote of the day Where are the adults in the room? Francesca Marano, a long-time contributor to WordPress, lambasts the platforms decision to require users to check a box reading Pineapple is delicious on pizza to log in, 404 Media reports. The big story Responsible AI has a burnout problem October 2022 Margaret Mitchell had been working at Google for two years before she realized she needed a break. Only after she spoke with a therapist did she understand the problem: she was burnt out. Mitchell, who now works as chief ethics scientist at the AI startup Hugging Face, is far from alone in her experience. Burnout is becoming increasingly common in responsible AI teams. All the practitioners MIT Technology Review interviewed spoke enthusiastically about their work: it is fueled by passion, a sense of urgency, and the satisfaction of building solutions for real problems. But that sense of mission can be overwhelming without the right support. Read the full story. Melissa Heikkil We can still have nice things A place for comfort, fun and distraction to brighten up your day. (Got any ideas? Drop me a line or skeet 'em at me.) + This timelapse of a pine tree growing from a tiny pinecone is pretty special + Shaboozeys A Bar Song (Tipsy) is one of 2024s biggest hits. But why has it struck such a chord?+ All hail Londons campest Christmas tree!+ Stay vigilant, Oregons googly eye bandit has struck again

Towana Looney, a 53-year-old woman from Alabama, has become the third living person to receive a kidney transplant from a gene-edited pig. Looney, who donated one of her kidneys to her mother back in 1999, developed kidney failure several years later following a pregnancy complication that caused high blood pressure. She started dialysis treatment in December of 2016 and was put on a waiting list for a kidney transplant soon after, in early 2017. But it was difficult to find a match. So Looneys doctors recommended the experimental pig organ as an alternative. After eight years on the waiting list, Looney was authorized to receive the kidney under the US Food and Drug Administrations expanded access program, which allows people with serious or life-threatening conditions to try experimental treatments. The pig in question was developed by Revivicor, a United Therapeutics company. The companys technique involves making 10 gene edits to a pig cell. The edits are made to prevent too much organ growth, curb inflammation, and, importantly, stop the recipients immune system from rejecting the organ. The edited pig cell is then placed into a pig egg cell that has had its nucleus removed, and the egg is transferred to the uterus of a sow, which eventually gives birth to a gene-edited piglet. JOE CARROTTA FOR NYU LANGONE HEALTH In theory, once the piglet has grown, its organs can be used for human transplantation. Pig organs are similar in size to human ones, after all. A few years ago, David Bennett Sr. became the first person to receive a heart transplant from such a pig. He died two months after the operation, and the heart was later found to have been infected with a pig virus. Richard Slayman was the first person to get a gene-edited pig kidney, which he received in early 2024. He died two months after his surgery, although the hospital treating him said in a statement that it had no indication that it was the result of his recent transplant. In April, Lisa Pisano was reported to be the second person to receive such an organ. Pisano also received a heart pump alongside her kidney transplant. Her kidney failed because of an inadequate blood supply and was removed the following month. She died in July. Looney received her pig kidney during a seven-hour operation that took place at NYU Langone Health in New York City on November 25. The surgery was led by Jayme Locke of the US Health Resources & Services Administration and Robert Montgomery of the NYU Langone Transplant Institute. Looney was discharged from the hospital 11 days after her surgery, to an apartment in New York City. Shell stay in New York for another three months so she can check in with doctors at the hospital for evaluations. Its a blessing, Looney said in a statement. I feel like Ive been given another chance at life. I cannot wait to be able to travel again and spend more quality time with my family and grandchildren. Looneys doctors are hopeful that her kidney will last longer than those of her predecessors. For a start, Looney was in better health to begin withshe had chronic kidney disease and required dialysis, but unlike previous recipients, she was not close to death, Montgomery said in a briefing. He and his colleagues plan to start clinical trials within the next year. There is a huge unmet need for organs. In the US alone, there more than 100,000 people are waiting for one, and 17 people on the waiting list die every day. Researchers hope that gene-edited animals might provide a new source of organs for such individuals. Revivicor isnt the only company working on this. Rival company eGenesis, which has a different approach to gene editing, has used CRISPR to create pigs with around 70 gene edits. Transplant is one of the few therapies that can cure a complex disease overnight, yet there are too few organs to provide a cure for all in need, Locke said in a statement. The thought that we may now have a solution to the organ shortage crisis for others who have languished on our waiting lists invokes the most welcome of feelings: pure joy! Today, Looney is the only person living with a pig organ. I am full of energy. I got an appetite Ive never had in eight years, she said at a briefing. I can put my hand on this kidney and feel it buzzing. This story has been updated with additional information after a press briefing.

Last week, this space was all about OpenAIs 12 days of shipmas. This week, the spotlight is on Google, which has been speeding toward the holiday by shipping or announcing its own flurry of products and updates. The combination of stuff here is pretty monumental, not just for a single company, but I think because it speaks to the power of the technology industryeven if it does trigger a personal desire that we could do more to harness that power and put it to more noble uses. To start, last week Google Introduced Veo, a new video generation model, and Imagen 3, a new version of its image generation model.Then on Monday, Google announced a breakthrough in quantum computing with its Willow chip. The company claims the new machine is capable of a standard benchmark computation in under five minutes that would take one of todays fastest supercomputers 10 septillion (that is, 1025) years. you may recall that MIT Technology Review covered some of the Willow work after researchers posted a paper preprint in August. But this week marked the big media splash. It was a stunning update that had Silicon Valley abuzz. (Seriously, I have never gotten so many quantum computing pitches as in the past few days.) Google followed this on Wednesday with even more gifts: a Gemini 2 release, a Project Astra update, and even more news about forthcoming agents called Mariner, an agent that can browse the web, and Jules, a coding assistant. First: Gemini 2. Its impressive, with a lot of performance updates. But I have frankly grown a little inured by language-model performance updates to the point of apathy. Or at least near-apathy. I want to see them do something. So for me, the cooler update was second on the list: Project Astra, which comes across like an AI from a futuristic movie set. Google first showed a demo of Astra back in May at its developer conference, and it was the talk of the show. But, since demos offer companies chances to show off products at their most polished, it can be hard to tell whats real and whats just staged for the audience. Still, when my colleague Will Douglas Heaven recently got to try it out himself, live and unscripted, it largely lived up to the hype. Although he found it glitchy, he noted that those glitches can be easily corrected. He called the experience stunning and said it could be generative AIs killer app.On top of all this, Will notes that this week Google DeepMind CEO (the companys AI division) Demis Hassabis was in Sweden to receive his Nobel Prize. And what did you do with your week? Making all this even more impressive, the advances represented in Willow, Gemini, Astra, and Veo are ones that just a few years ago many, many people would have said were not possibleor at least not in this timeframe.A popular knock on the tech industry is that it has a tendency to over-promise and under-deliver. The phone in your pocket gives the lie to this. So too do the rides I took in Waymos self-driving cars this week. (Both of which arrived faster than Ubers estimated wait time. And honestly its not been that long since the mere ability to summon an Uber was cool!) And while quantum has a long way to go, the Willow announcement seems like an exceptional advance; if not a tipping point exactly, then at least a real waypoint on a long road. (For what its worth, Im still not totally sold on chatbots. They do offer novel ways of interacting with computers, and have revolutionized information retrieval. But whether they are beneficial for humanityespecially given energy debts, the use of copyrighted material in their training data, their perhaps insurmountable tendency to hallucinate, etc.is debatable, and certainly is being debated. But Im pretty floored by this weeks announcements from Google, as well as OpenAIfull stop.) And for all the necessary and overdue talk about reining in the power of Big Tech, the ability to hit significant new milestones on so many different fronts all at once is something that only a company with the resources of a Google (or Apple or Microsoft or Amazon or Meta or Baidu or whichever other behemoth) can do.All this said, I dont want us to buy more gadgets or spend more time looking at our screens. I dont want us to become more isolated physically, socializing with others only via our electronic devices. I dont want us to fill the air with carbon or our soil with e-waste. I do not think these things should be the price we pay to drive progress forward. Its indisputable that humanity would be better served if more of the tech industry was focused on ending poverty and hunger and disease and war. Yet every once in a while, in the ever-rising tide of hype and nonsense that pumps out of Silicon Valley, epitomized by the AI gold rush of the past couple of years, there are moments that make me sit back in awe and amazement at what people can achieve, and in which I become hopeful about our ability to actually solve our larger problemsif only because we can solve so many other dumber, but incredibly complicated ones. This week was one of those times for me. Now read the rest of The Debrief The News Robotaxi adoptionis hitting a tipping point. But also,GM is shutting down its Cruise robotaxi division. Hereshow to use OpenAIs new video editing toolSora. Blueskyhas an impersonator problem. The AI hype machine iscoming under government scrutiny. The Chat Every week, I talk to one of MIT Technology Reviews journalists to go behind the scenes of a story they are working on. This week, I hit up James ODonnell, who covers AI and hardware, about his story on how the startup defense contractorAnduril is bringing AI to the battlefield. Mat:James, you got a pretty up close look at something most people probably havent even thought about yet, which is how the future of AI-assisted warfare might look. What did you learn on that trip that you think will surprise people? James:Two things stand out. One, I think people would be surprised by the gulf between how technology has developed for the last 15 years for consumers versus the military. For consumers, weve gotten phones, computers, smart TVs and other technologies that generally do a pretty good job of talking to each other and sharing our data, even though theyre made by dozens of different manufacturers. Its called the internet of things. In the military, technology has developed in exactly the opposite way, and its putting them in a crisis. They have stealth aircraft all over the world, but communicating about a drone threat might be done with Powerpoints and a chat service reminiscent of AOL Instant Messenger. The second is just how much the Pentagon is now looking to AI to change all of this. New initiatives have surged in the current AI boom. They are spending on training new AI models to better detect threats, autonomous fighter jets, and intelligence platforms that use AI to find pertinent information. What I saw at Andurils test site in California is also a key piece of that. Using AI to connect to and control lots of different pieces of hardware, like drones and cameras and submarines, from a single platform. The amount being invested in AI is much smaller than for aircraft carriers and jets, but its growing. Mat:I was talking with a different startup defense contractor recently, who was talking to me about the difficulty of getting all these increasingly autonomous devices on the battlefield talking to each other in a coordinated way. Like Anduril, he was making the case that this has to be done at the edge, and that there is too much happening for human decision making to process. Do you think thats true? Why is that? James:So many in the defense space have pointed to the war in Ukraine as a sign that warfare is changing. Drones are cheaper and more capable than they ever were in the wars in the Middle East. Its why the Pentagon is spending $1 billion on the Replicator initiative to fieldthousands of cheap dronesby 2025. Its also looking to field more underwater drones as it plans for scenarios in which China may invade Taiwan. Once you get these systems, though, the problem is having all the devices communicate with one another securely. You need to play Air Traffic Control at the same time that youre pulling in satellite imagery and intelligence information, all in environments where communication links are vulnerable to attacks. Mat:I guess I still have a mental image of a control room somewhere, like you might see inDr. StrangeloveorWar Games(orStar Warsfor that matter) with a handful of humans directing things. Are those days over? James:I think a couple things will change. One, a single person in that control room will be responsible for a lot more than they are now. Rather than running just one camera or drone system manually, theyll command software that does it for them, for lots of different devices. The idea that the defense tech sector is pushing is to take them out of the mundane tasksrotating a camera around to look for threatsand instead put them in the drivers seat for decisions that only humans, not machines, can make. Mat:I know that critics of the industry push back on the idea of AI being empowered to make battlefield decisions, particularly when it comes to life and death, but it seems to me that we are increasingly creeping toward that and it seems perhaps inevitable. Whats your sense? James:This is painting with broad strokes, but I think the debates about military AI fall along similar lines to what we see for autonomous vehicles. You have proponents saying that driving is not a thing humans are particularly good at, and when they make mistakes, it takes lives. Others might agree conceptually, but debate at what point its appropriate to fully adopt fallible self-driving technology in the real world. How much better does it have to be than humans? In the military, the stakes are higher. Theres no question that AI is increasingly being used to sort through and surface information to decision-makers. Its finding patterns in data, translating information, and identifying possible threats. Proponents are outspoken that that will make warfare more precise and reduce casualties. What critics are concerned about is how far across that decision-making pipeline AI is going, and how much there is human oversight. I think where it leaves me is wanting transparency. When AI systems make mistakes, just like when human military commanders make mistakes, I think we deserve to know, and that transparency does not have to compromise national security. It tookyearsfor reporter Azmat Khan to piece together the mistakes made during drone strikes in the Middle East, because agencies were not forthcoming. That obfuscation absolutely cannot be the norm as we enter the age of military AI. Mat:Finally, did you have a chance to hit an In-N-Out burger while you were in California? James:Normally In-N-Out is a requisite stop for me in California, but ahead of my trip I heard lots of good things about the burgers at The Apple Pan in West LA, so I went there. To be honest, the fries were better, but for the burger I have to hand it to In-N-Out. The Recommendation A few weeks ago I suggestedCa7riel and Paco Amorosos appearance on NPR Tiny Desk. At the risk of this space becoming a Tiny Desk stan account, Im back again with another. I was completely floored byDoechiis Tiny Desk appearance last week. Its so full of talent and joy and style and power. I came away completely inspired and have basically had her music on repeat in Spotify ever since. If you are already a fan of her recorded music, you will love her live. If shes new to you, well, youre welcome. Go check it out. Oh, and dont worry: Im not planning to recommendBillie Eilishs new Tiny Desk concertin next weeks newsletter. Mostly because Im doing so now.

AIs emissions are about to skyrocket even further Its no secret that the current AI boom is using up immense amounts of energy. Now we have a better idea of how much. A new paper, from a team at the Harvard T.H. Chan School of Public Health, examined 78% of all data centers in the country in the US. These facilitiesessentially buildings filled to the brim with rows of serversare where AI models get trained, and they also get pinged every time we send a request through models like ChatGPT. They require huge amounts of energy both to power the servers and to keep them cool. Since 2018, carbon emissions from data centers in the US have tripled.Its difficult to put a number on how much AI in particular is responsible for this surge. But AIs share is certainly growing rapidly as nearly every segment of the economy attempts to adopt the technology. Read the full story. Googles big week was a flex for the power of big tech Google has been speeding toward the holiday by shipping or announcing a flurry of products and updates. The combination of stuff here is pretty monumental, not just for a single company, but I think because it speaks to the power of the technology industryeven if it does trigger a personal desire that we could do more to harness that power and put it to more noble uses.Read more here. This story originally appeared in The Debrief with Mat Honan, our weekly take on whats really going on behind the biggest tech headlines. The story is subscriber-only sonab a subscriptiontoo, if you havent already! Or you cansign upto the newsletter for free to get the next edition in your inbox on Friday. The must-reads Ive combed the internet to find you todays most fun/important/scary/fascinating stories about technology. 1 Mysterious drones have been spotted along the US east coast People are getting a bit freaked out, to say the least. (BBC) Although sometimes theyre just small planes, authorities say.(Wired) Trump says they should be shot down.(Politico) 2 TikTok could be gone from app stores by January 19 Last week, a US appeals court upheld a law forcing Bytedance to divest. (Reuters) The rationale behind the ban could open the door to other regulations that suppress speech.(Atlantic) Influencers are putting together their post-TikTok plans.(Business Insider) The long-shot plan to save TikTok. (Verge) The depressing truth about the coming ban.(MIT Technology Review) 3 Authorities in Serbia are using phone-cracking tools to install spyware Activists and journalists found their phone had been tampered with after a run-in with police. (404 Media) 4 Cellphone videos are fueling violence inside US schools Students are using phones to arrange, provoke and capture brawls in the corridors. (NYT) 5 AI search startup Perplexity says it will generate $10.5 million a month next year Its in talks to raise money at a $9 billion valuation. (The Information) AI search could break the web. (MIT Technology Review) 6 How Musks partnership with Trump could influence science Even if he cant cut as much as hed like, he still stands to make big changes. (Nature) Is deleting the IRS his worst idea yet?(Washington Post) The top cybersecurity agency is bracing for Trump. (Wired) Trumps win is a huge loss for the climate.(MIT Technology Review) 7 AI firms will scour the globe looking for cheap energy Low-cost power is an absolute priority. (Wired) Its an insatiably hungry industry.(Bloomberg) 8 Anthropics Claude is winning the chatbot battle for tech insiders Its not as big as ChatGPT, but it's got a special something that people like. (NYT) A new Character.ai chatbot for teens will no longer talk romance. (Verge) How to trust what a chatbot says.(MIT Technology Review) 9 The reaction to the UnitedHealthcare CEOs murder could prompt a reckoning Healthcares algorithmic decision-making turns us into numbers on a spreadsheets. (Vanity Fair) Luigi Mangione has to mean something. (Atlantic) 10 How Chinas satellite megaprojects are challenging Starlink Between them, Qianfan, Guo Wang and Honghu-3 could have as many satellites.(CNBC) Quote of the day Weve achieved peak data and therell be no more. OpenAIs cofounder and former chief scientist, Ilya Sutskever, tells the NeurIPS conference that the way AI models will be trained will have to change. The big story How to stop a state from sinkingApril 2024 In a 10-month span between 2020 and 2021, southwest Louisiana saw five climate-related disasters, including two destructive hurricanes. As if that wasnt bad enough, more storms are coming, and many areas are not prepared. But some government officials and state engineers are hoping there is an alternative: elevation. The $6.8 billion Southwest Coastal Louisiana Project is betting that raising residences by a few feet, coupled with extensive work to restore coastal boundary lands, will keep Louisianans in their communities. Ultimately, its something of a last-ditch effort to preserve this slice of coastline, even as some locals pick up and move inland and as formal plans for managed retreat become more popular in climate-vulnerable areas across the country and the rest of the world.Read the full story. Xander Peters We can still have nice things A place for comfort, fun and distraction to brighten up your day. (Got any ideas?Drop me a lineortweet 'em at me.)+ How to make the most of yourjigsaw puzzlestry them on hard mode. +Mr Tickleis a maniac who needs to be stopped.+ Asong about Christmasthat probably many of us can relate to, if were honest. + If the original Home Alone was wince-inducing in terms of injuries, thesequelis even more excruciating.+ The best crispy roast potatoes ever?Ill let you be the judge.

Last week, this space was all about OpenAIs 12 days of shipmas. This week, the spotlight is on Google, which has been speeding toward the holiday by shipping or announcing its own flurry of products and updates. The combination of stuff here is pretty monumental, not just for a single company, but I think because it speaks to the power of the technology industryeven if it does trigger a personal desire that we could do more to harness that power and put it to more noble uses. To start, last week Google Introduced Veo, a new video generation model, and Imagen 3, a new version of its image generation model.Then on Monday, Google announced a breakthrough in quantum computing with its Willow chip. The company claims the new machine is capable of a standard benchmark computation in under five minutes that would take one of todays fastest supercomputers 10 septillion (that is, 1025) years. you may recall that MIT Technology Review covered some of the Willow work after researchers posted a paper preprint in August. But this week marked the big media splash. It was a stunning update that had Silicon Valley abuzz. (Seriously, I have never gotten so many quantum computing pitches as in the past few days.) Google followed this on Wednesday with even more gifts: a Gemini 2 release, a Project Astra update, and even more news about forthcoming agents called Mariner, an agent that can browse the web, and Jules, a coding assistant. First: Gemini 2. Its impressive, with a lot of performance updates. But I have frankly grown a little inured by language-model performance updates to the point of apathy. Or at least near-apathy. I want to see them do something. So for me, the cooler update was second on the list: Project Astra, which comes across like an AI from a futuristic movie set. Google first showed a demo of Astra back in May at its developer conference, and it was the talk of the show. But, since demos offer companies chances to show off products at their most polished, it can be hard to tell whats real and whats just staged for the audience. Still, when my colleague Will Douglas Heaven recently got to try it out himself, live and unscripted, it largely lived up to the hype. Although he found it glitchy, he noted that those glitches can be easily corrected. He called the experience stunning and said it could be generative AIs killer app.On top of all this, Will notes that this week Google DeepMind CEO (the companys AI division) Demis Hassabis was in Sweden to receive his Nobel Prize. And what did you do with your week? Making all this even more impressive, the advances represented in Willow, Gemini, Astra, and Veo are ones that just a few years ago many, many people would have said were not possibleor at least not in this timeframe.A popular knock on the tech industry is that it has a tendency to over-promise and under-deliver. The phone in your pocket gives the lie to this. So too do the rides I took in Waymos self-driving cars this week. (Both of which arrived faster than Ubers estimated wait time. And honestly its not been that long since the mere ability to summon an Uber was cool!) And while quantum has a long way to go, the Willow announcement seems like an exceptional advance; if not a tipping point exactly, then at least a real waypoint on a long road. (For what its worth, Im still not totally sold on chatbots. They do offer novel ways of interacting with computers, and have revolutionized information retrieval. But whether they are beneficial for humanityespecially given energy debts, the use of copyrighted material in their training data, their perhaps insurmountable tendency to hallucinate, etc.is debatable, and certainly is being debated. But Im pretty floored by this weeks announcements from Google, as well as OpenAIfull stop.) And for all the necessary and overdue talk about reining in the power of Big Tech, the ability to hit significant new milestones on so many different fronts all at once is something that only a company with the resources of a Google (or Apple or Microsoft or Amazon or Meta or Baidu or whichever other behemoth) can do.All this said, I dont want us to buy more gadgets or spend more time looking at our screens. I dont want us to become more isolated physically, socializing with others only via our electronic devices. I dont want us to fill the air with carbon or our soil with e-waste. I do not think these things should be the price we pay to drive progress forward. Its indisputable that humanity would be better served if more of the tech industry was focused on ending poverty and hunger and disease and war. Yet every once in a while, in the ever-rising tide of hype and nonsense that pumps out of Silicon Valley, epitomized by the AI gold rush of the past couple of years, there are moments that make me sit back in awe and amazement at what people can achieve, and in which I become hopeful about our ability to actually solve our larger problemsif only because we can solve so many other dumber, but incredibly complicated ones. This week was one of those times for me. Now read the rest of The Debrief The News Robotaxi adoptionis hitting a tipping point. But also,GM is shutting down its Cruise robotaxi division. Hereshow to use OpenAIs new video editing toolSora. Blueskyhas an impersonator problem. The AI hype machine iscoming under government scrutiny. The Chat Every week, I talk to one of MIT Technology Reviews journalists to go behind the scenes of a story they are working on. This week, I hit up James ODonnell, who covers AI and hardware, about his story on how the startup defense contractorAnduril is bringing AI to the battlefield. Mat:James, you got a pretty up close look at something most people probably havent even thought about yet, which is how the future of AI-assisted warfare might look. What did you learn on that trip that you think will surprise people? James:Two things stand out. One, I think people would be surprised by the gulf between how technology has developed for the last 15 years for consumers versus the military. For consumers, weve gotten phones, computers, smart TVs and other technologies that generally do a pretty good job of talking to each other and sharing our data, even though theyre made by dozens of different manufacturers. Its called the internet of things. In the military, technology has developed in exactly the opposite way, and its putting them in a crisis. They have stealth aircraft all over the world, but communicating about a drone threat might be done with Powerpoints and a chat service reminiscent of AOL Instant Messenger. The second is just how much the Pentagon is now looking to AI to change all of this. New initiatives have surged in the current AI boom. They are spending on training new AI models to better detect threats, autonomous fighter jets, and intelligence platforms that use AI to find pertinent information. What I saw at Andurils test site in California is also a key piece of that. Using AI to connect to and control lots of different pieces of hardware, like drones and cameras and submarines, from a single platform. The amount being invested in AI is much smaller than for aircraft carriers and jets, but its growing. Mat:I was talking with a different startup defense contractor recently, who was talking to me about the difficulty of getting all these increasingly autonomous devices on the battlefield talking to each other in a coordinated way. Like Anduril, he was making the case that this has to be done at the edge, and that there is too much happening for human decision making to process. Do you think thats true? Why is that? James:So many in the defense space have pointed to the war in Ukraine as a sign that warfare is changing. Drones are cheaper and more capable than they ever were in the wars in the Middle East. Its why the Pentagon is spending $1 billion on the Replicator initiative to fieldthousands of cheap dronesby 2025. Its also looking to field more underwater drones as it plans for scenarios in which China may invade Taiwan. Once you get these systems, though, the problem is having all the devices communicate with one another securely. You need to play Air Traffic Control at the same time that youre pulling in satellite imagery and intelligence information, all in environments where communication links are vulnerable to attacks. Mat:I guess I still have a mental image of a control room somewhere, like you might see inDr. StrangeloveorWar Games(orStar Warsfor that matter) with a handful of humans directing things. Are those days over? James:I think a couple things will change. One, a single person in that control room will be responsible for a lot more than they are now. Rather than running just one camera or drone system manually, theyll command software that does it for them, for lots of different devices. The idea that the defense tech sector is pushing is to take them out of the mundane tasksrotating a camera around to look for threatsand instead put them in the drivers seat for decisions that only humans, not machines, can make. Mat:I know that critics of the industry push back on the idea of AI being empowered to make battlefield decisions, particularly when it comes to life and death, but it seems to me that we are increasingly creeping toward that and it seems perhaps inevitable. Whats your sense? James:This is painting with broad strokes, but I think the debates about military AI fall along similar lines to what we see for autonomous vehicles. You have proponents saying that driving is not a thing humans are particularly good at, and when they make mistakes, it takes lives. Others might agree conceptually, but debate at what point its appropriate to fully adopt fallible self-driving technology in the real world. How much better does it have to be than humans? In the military, the stakes are higher. Theres no question that AI is increasingly being used to sort through and surface information to decision-makers. Its finding patterns in data, translating information, and identifying possible threats. Proponents are outspoken that that will make warfare more precise and reduce casualties. What critics are concerned about is how far across that decision-making pipeline AI is going, and how much there is human oversight. I think where it leaves me is wanting transparency. When AI systems make mistakes, just like when human military commanders make mistakes, I think we deserve to know, and that transparency does not have to compromise national security. It tookyearsfor reporter Azmat Khan to piece together the mistakes made during drone strikes in the Middle East, because agencies were not forthcoming. That obfuscation absolutely cannot be the norm as we enter the age of military AI. Mat:Finally, did you have a chance to hit an In-N-Out burger while you were in California? James:Normally In-N-Out is a requisite stop for me in California, but ahead of my trip I heard lots of good things about the burgers at The Apple Pan in West LA, so I went there. To be honest, the fries were better, but for the burger I have to hand it to In-N-Out. The Recommendation A few weeks ago I suggestedCa7riel and Paco Amorosos appearance on NPR Tiny Desk. At the risk of this space becoming a Tiny Desk stan account, Im back again with another. I was completely floored byDoechiis Tiny Desk appearance last week. Its so full of talent and joy and style and power. I came away completely inspired and have basically had her music on repeat in Spotify ever since. If you are already a fan of her recorded music, you will love her live. If shes new to you, well, youre welcome. Go check it out. Oh, and dont worry: Im not planning to recommendBillie Eilishs new Tiny Desk concertin next weeks newsletter. Mostly because Im doing so now.

Like many others, I recently fled the social media platform X for Bluesky. In the process, I started following many of the people I followed on X. On Thanksgiving, I was delighted to see a private message from a fellow AI reporter, Will Knight from Wired. Or at least thats who I thought I was talking to. I became suspicious when the person claiming to be Knight mentioned being from Miami, when Knight is, in fact, from the UK. The account handle was almost identical to the real Will Knights handle, and the profile used his profile photo. Then more messages started to appear. Paris Marx, a prominent tech critic, slid into my DMs to ask me how I was doing. Things are going splendid over here, he replied to me. Then things got suspicious again. How are your trades going? fake-Marx asked me. This account was far more sophisticated than Knights; it had meticulously copied every single tweet and retweet from Marxs real page over the past few weeks. Both accounts were eventually deleted, but not before trying to get me to set up a crypto wallet and a cloud mining pool account. Knight and Marx confirmed to us that these accounts did not belong to them, and that they have been fighting impersonator accounts of themselves for weeks. They are not the only ones. The New York Times tech journalist Sheera Frankel and Molly White, a researcher and cryptocurrency critic, have also experienced people impersonating them on Bluesky, most likely to scam people. This tracks with research from Alexios Mantzarlis, the director of the Security, Trust, and Safety Initiative at Cornell Tech, who manually went through the top 500 Bluesky users by follower count and found that of the 305 accounts belonging to a named person, at least 74 had been impersonated by at least one other account. The platform has had to suddenly cater to an influx of millions of new users in recent months as people leave X in protest of Elon Musks takeover of the platform. Its user base has more than doubled since September, from 10 million users to over 20 million. This sudden wave of new usersand the inevitable scammersmeans Bluesky is still playing catch-up, says White. These accounts block me as soon as theyre created, so I dont initially see them, Marx says. Both Marx and White describe a frustrating pattern: When one account is taken down, another one pops up soon after. White says she had experienced a similar phenomenon on X and TikTok too. A way to prove that people are who they say they are would help. Before Musk took the reins of the platform, employees at X, previously known as Twitter, verified users such as journalists and politicians, and gave them a blue tick next to their handles so people knew they were dealing with credible news sources. After Musk took over, he scrapped the old verification system and offered blue ticks to all paying customers. The ongoing crypto-impersonation scams have raised calls for Bluesky to initiate something similar to Twitters original verification program. Some users, such as the investigative journalist Hunter Walker, have set up their own initiatives to verify journalists. However, users are currently limited in the ways they can verify themselves on the platform. By default, usernames on Bluesky end with the suffix bsky.social. The platform recommends that news organizations and high-profile people verify their identities by setting up their own websites as their usernames. For example, US senators have verified their accounts with the suffix senate.gov. But this technique isnt foolproof. For one, it doesnt actually verify peoples identityonly their affiliation with a particular website. Bluesky did not respond to MIT Technology Reviews requests for comment, but the companys safety team posted that the platform had updated its impersonation policy to be more aggressive and would remove impersonation and handle-squatting accounts. The company says it has also quadrupled its moderation team to take action on impersonation reports more quickly. But it seems to be struggling to keep up. We still have a large backlog of moderation reports due to the influx of new users as we shared previously, though we are making progress, the company continued. Blueskys decentralized nature makes kicking out impersonators a trickier problem to solve. Competitors such as X and Threads rely on centralized teams within the company who moderate unwanted content and behavior, such as impersonation. But Bluesky is built on the AT Protocol, a decentralized, open-source technology, which allows users more control over what kind of content they see and enables them to build communities around particular content. Most people sign up to Bluesky Social, the main social network, whose community guidelines ban impersonation. However, Bluesky Social is just one of the services or clients that people can use, and other services have their own moderation practices and terms. This approach means that until now, Bluesky itself hasnt needed an army of content moderators to weed out unwanted behaviors because it relies on this community-led approach, says Wayne Chang, the founder and CEO of SpruceID, a digital identity company. That might have to change. In order to make these apps work at all, you need some level of centralization, says Chang. Despite community guidelines, its hard to stop people from creating impersonation accounts, and Bluesky is engaged in a cat-and-mouse game trying to take suspicious accounts down. Cracking down on a problem such as impersonation is important because it poses a serious problem for the credibility of Bluesky, says Chang. Its a legitimate complaint as a Bluesky user that Hey, all those scammers are basically harassing me. You want your brand to be tarnished? Or is there something we can do about this? he says. A fix for this is urgently needed, because attackers might abuse Blueskys open-source code to create spam and disinformation campaigns at a much larger scale, says Francesco Pierri, an assistant professor at Politecnico di Milano who has researched Bluesky. His team found that the platform has seen a rise in suspicious accounts since it was made open to the public earlier this year. Bluesky acknowledges that its current practices are not enough. In a post, the company said it has received feedback that users want more ways to confirm their identities beyond domain verification, and it is exploring additional options to enhance account verification. In a livestream at the end of November, Bluesky CEO Jay Graber said the platform was considering becoming a verification provider, but because of its decentralized approach it would also allow others to offer their own user verification services. And [users] can choose to trust usthe Bluesky teams verificationor they could do their own. Or other people could do their own, Graber said. But at least Bluesky seems to have some willingness to actually moderate content on the platform, says White. I would love to see something a little bit more proactive that didnt require me to do all of this reporting, she adds. As for Marx, I just hope that no one truly falls for it and gets tricked into crypto scams, he says.

The internet loves a good neologism, especially if it can capture a purported vibe shift or explain a new trend. In 2013, the columnist Adrian Wooldridge coined a word that eventually did both. Writing for the Economist, he warned of the coming techlash, a revolt against Silicon Valleys rich and powerful fueled by the publics growing realization that these sovereigns of cyberspace werent the benevolent bright-future bringers they claimed to be. While Wooldridge didnt say precisely when this techlash would arrive, its clear today that a dramatic shift in public opinion toward Big Tech and its leaders did in fact happenand is arguably still happening. Say what you will about the legions of Elon Musk acolytes on X, but if an industry and its executives can bring together the likes of Elizabeth Warren and Lindsey Graham in shared condemnation, its definitely not winning many popularity contests. To be clear, there have always been critics of Silicon Valleys very real excesses and abuses. But for the better part of the last two decades, many of those voices of dissent were either written off as hopeless Luddites and haters of progress or drowned out by a louder and far more numerous group of techno-optimists. Today, those same critics (along with many new ones) have entered the fray once more, rearmed with popular Substacks, media columns, andincreasinglybook deals. Two of the more recent additions to the flourishing techlash genreRob Lalkas The Venture Alchemists: How Big Tech Turned Profits into Power and Marietje Schaakes The Tech Coup: How to Save Democracy from Silicon Valleyserve as excellent reminders of why it started in the first place. Together, the books chronicle the rise of an industry that is increasingly using its unprecedented wealth and power to undermine democracy, and they outline what we can do to start taking some of that power back. Lalka is a business professor at Tulane University, and The Venture Alchemists focuses on how a small group of entrepreneurs managed to transmute a handful of novel ideas and big bets into unprecedented wealth and influence. While the names of these demigods of disruption will likely be familiar to anyone with an internet connection and a passing interest in Silicon Valley, Lalka also begins his book with a page featuring their nine (mostly) young, (mostly) smiling faces. There are photos of the famous founders Mark Zuckerberg, Larry Page, and Sergey Brin; the VC funders Keith Rabois, Peter Thiel, and David Sacks; and a more motley trio made up of the disgraced former Uber CEO Travis Kalanick, the ardent eugenicist and reputed father of Silicon Valley Bill Shockley (who, it should be noted, died in 1989), and a former VC and the future vice president of the United States, JD Vance. To his credit, Lalka takes this medley of tech titans and uses their origin stories and interrelationships to explain how the so-called Silicon Valley mindset (mind virus?) became not just a fixture in Californias Santa Clara County but also the preeminent way of thinking about success and innovation across America. This approach to doing business, usually cloaked in a barrage of cringey innovation-speakdisrupt or be disrupted, move fast and break things, better to ask for forgiveness than permissioncan often mask a darker, more authoritarian ethos, according to Lalka. One of the nine entrepreneurs in the book, Peter Thiel, has written that I no longer believe that freedom and democracy are compatible and that competition [in business] is for losers. Many of the others think that all technological progress is inherently good and should be pursued at any cost and for its own sake. A few also believe that privacy is an antiquated concepteven an illusionand that their companies should be free to hoard and profit off our personal data. Most of all, though, Lalka argues, these men believe that their newfound power should be unconstrained by governments, regulators, or anyone else who might have the gall to impose some limitations. Where exactly did these beliefs come from? Lalka points to people like the late free-market economist Milton Friedman, who famously asserted that a companys only social responsibility is to increase profits, as well as to Ayn Rand, the author, philosopher, and hero to misunderstood teenage boys everywhere who tried to turn selfishness into a virtue. The Venture Alchemists: How Big Tech Turned Profits into PowerRob LalkaCOLUMBIA BUSINESS SCHOOL PUBLISHING, 2024 Its a somewhat reductive and not altogether original explanation of Silicon Valleys libertarian inclinations. What ultimately matters, though, is that many of these values were subsequently encoded into the DNA of the companies these men founded and fundedcompanies that today shape how we communicate with one another, how we share and consume news, and even how we think about our place in the world. The Venture Alchemists is strongest when its describing the early-stage antics and on-campus controversies that shaped these young entrepreneurs or, in many cases, simply reveal who theyve always been. Lalka is a thorough and tenacious researcher, as the books 135 pages of endnotes suggest. And while nearly all these stories have been told before in other books and articles, he still manages to provide new perspectives and insights from sources like college newspapers and leaked documents. One thing the book is particularly effective at is deflating the myth that these entrepreneurs were somehow gifted seers of (and investors in) a future the rest of us simply couldnt comprehend or predict. Sure, someone like Thiel made what turned out to be a savvy investment in Facebook early on, but he also made some very costly mistakes with that stake. As Lalka points out, Thiels Founders Fund dumped tens of millions of shares shortly after Facebook went public, and Thiel himself went from owning 2.5% of the company in 2012 to 0.000004% less than a decade later (around the same time Facebook hit its trillion-dollar valuation). Throw in his objectively terrible wagers in 2008, 2009, and beyond, when he effectively shorted what turned out to be one of the longest bull markets in world history, and you get the impression hes less oracle and more ideologue who happened to take some big risks that paid off. One of Lalkas favorite mantras throughout The Venture Alchemists is that words matter. Indeed, he uses a lot of these entrepreneurs own words to expose their hypocrisy, bullying, juvenile contrarianism, casual racism, andyesoutright greed and self-interest. It is not a flattering picture, to say the least. Unfortunately, instead of simply letting those words and deeds speak for themselves, Lalka often feels the need to interject with his own, frequently enjoining readers against finger-pointing or judging these men too harshly even after hes chronicled their many transgressions. Whether this is done to try to convey some sense of objectivity or simply to remind readers that these entrepreneurs are complex and complicated men making difficult decisions, it doesnt work. At all. For one thing, Lalka clearly has his own strong opinions about the behavior of these entrepreneursopinions he doesnt try to disguise. At one point in the book he suggests that Kalanicks alpha-male, dominance-at-any-cost approach to running Uber is almost, but not quite like rape, which is maybe not the comparison youd make if you wanted to seem like an arbiter of impartiality. And if he truly wants readers to come to a different conclusion about these men, he certainly doesnt provide many reasons for doing so. Simply telling us to judge less, and discern more seems worse than a cop-out. It comes across as almost, but not quite like victim-blamingas if were somehow just as culpable as they are for using their platforms and buying into their self-mythologizing. In many ways, Silicon Valley has become the antithesis of what its early pioneers set out to be. Marietje Schaake Equally frustrating is the crescendo of empty platitudes that ends the book. The technologies of the future must be pursued thoughtfully, ethically, and cautiously, Lalka says after spending 313 pages showing readers how these entrepreneurs have willfully ignored all three adverbs. What theyve built instead are massive wealth-creation machines that divide, distract, and spy on us. Maybe its just me, but that kind of behavior seems ripe not only for judgment, but also for action. So what exactly do you do with a group of men seemingly incapable of serious self-reflectionmen who believe unequivocally in their own greatness and who are comfortable making decisions on behalf of hundreds of millions of people who did not elect them, and who do not necessarily share their values? You regulate them, of course. Or at least you regulate the companies they run and fund. In Marietje Schaakes The Tech Coup, readers are presented with a road map for how such regulation might take shape, along with an eye-opening account of just how much power has already been ceded to these corporations over the past 20 years. There are companies like NSO Group, whose powerful Pegasus spyware tool has been sold to autocrats, who have in turn used it to crack down on dissent and monitor their critics. Billionaires are now effectively making national security decisions on behalf of the United States and using their social media companies to push right-wing agitprop and conspiracy theories, as Musk does with his Starlink satellites and X. Ride-sharing companies use their own apps as propaganda tools and funnel hundreds of millions of dollars into ballot initiatives to undo laws they dont like. The list goes on and on. According to Schaake, this outsize and largely unaccountable power is changing the fundamental ways that democracy works in the United States. In many ways, Silicon Valley has become the antithesis of what its early pioneers set out to be: from dismissing government to literally taking on equivalent functions; from lauding freedom of speech to becoming curators and speech regulators; and from criticizing government overreach and abuse to accelerating it through spyware tools and opaque algorithms, she writes. Schaake, whos a former member of the European Parliament and the current international policy director at Stanford Universitys Cyber Policy Center, is in many ways the perfect chronicler of Big Techs power grab. Beyond her clear expertise in the realms of governance and technology, shes also Dutch, which makes her immune to the distinctly American disease that seems to equate extreme wealth, and the power that comes with it, with virtue and intelligence. This resistance to the various reality-distortion fields emanating from Silicon Valley plays a pivotal role in her ability to see through the many justifications and self-serving solutions that come from tech leaders themselves. Schaake understands, for instance, that when someone like OpenAIs Sam Altman gets in front of Congress and begs for AI regulation, what hes really doing is asking Congress to create a kind of regulatory moat between his company and any other startups that might threaten it, not acting out of some genuine desire for accountability or governmental guardrails. The Tech Coup:How to Save Democracyfrom Silicon ValleyMarietje SchaakePRINCETON UNIVERSITY PRESS, 2024 Like Shoshana Zuboff, the author of The Age of Surveillance Capitalism, Schaake believes that the digital should live within democracys housethat is, technologies should be developed within the framework of democracy, not the other way around. To accomplish this realignment, she offers a range of solutions, from banning what she sees as clearly antidemocratic technologies (like face-recognition software and other spyware tools) to creating independent teams of expert advisors to members of Congress (who are often clearly out of their depth when attempting to understand technologies and business models). Predictably, all this renewed interest in regulation has inspired its own backlash in recent yearsa kind of tech revanchism, to borrow a phrase from the journalist James Hennessy. In addition to familiar attacks, such as trying to paint supporters of the techlash as somehow being antitechnology (theyre not), companies are also spending massive amounts of money to bolster their lobbying efforts. Some venture capitalists, like LinkedIn cofounder Reid Hoffman, who made big donations to the Kamala Harris presidential campaign, wanted to evict Federal Trade Commission chair Lina Khan, claiming that regulation is killing innovation (it isnt) and removing the incentives to start a company (its not). And then of course theres Musk, who now seems to be in a league of his own when it comes to how much influence he may exert over Donald Trump and the government that his companies have valuable contracts with. What all these claims of victimization and subsequent efforts to buy their way out of regulatory oversight miss is that theres actually a vast and fertile middle ground between simple techno-optimism and techno-skepticism. As the New Yorker contributor Cal Newport and others have noted, its entirely possible to support innovations that can significantly improve our lives without accepting that every popular invention is good or inevitable. Regulating Big Tech will be a crucial part of leveling the playing field and ensuring that the basic duties of a democracy can be fulfilled. But as both Lalka and Schaake suggest, another battle may prove even more difficult and contentious. This one involves undoing the flawed logic and cynical, self-serving philosophies that have led us to the point where we are now. What if we admitted that constant bacchanals of disruption are in fact not all that good for our planet or our brains? What if, instead of creative destruction, we started fetishizing stability, and in lieu of putting dents in the universe, we refocused our efforts on fixing whats already broken? What ifand hear me outwe admitted that technology might not be the solution to every problem we face as a society, and that while innovation and technological change can undoubtedly yield societal benefits, they dont have to be the only measures of economic success and quality of life? When ideas like these start to sound less like radical concepts and more like common sense, well know the techlash has finally achieved something truly revolutionary. Bryan Gardiner is a writer based in Oakland, California.

This is today's edition ofThe Download,our weekday newsletter that provides a daily dose of what's going on in the world of technology. How Silicon Valley is disrupting democracy The internet loves a good neologism, especially if it can capture a purported vibe shift or explain a new trend. In 2013, the columnist Adrian Wooldridge coined a word that eventually did both. Writing for the Economist, he warned of the coming techlash, a revolt against Silicon Valleys rich and powerful fueled by the publics growing realization that these sovereigns of cyberspace werent the benevolent bright-future bringers they claimed to be. While Wooldridge didnt say precisely when this techlash would arrive, its clear today that a dramatic shift in public opinion toward Big Tech and its leaders did in fact happenand is arguably still happening. Two new books serve as excellent reminders of why it started in the first place. Together, they chronicle the rise of an industry that is increasingly using its unprecedented wealth and power to undermine democracy, and they outline what we can do to start taking some of that power back. Read the full story.Bryan Gardiner This story is from the forthcoming magazine edition of MIT Technology Review, set to go live on January 6its all about the exciting breakthroughs happening in the world right now. If you dont already, subscribe to receive a copy. The must-reads Ive combed the internet to find you todays most fun/important/scary/fascinating stories about technology. 1 Google has unveiled a new headset and smart glasses OS Android XR gives wearers hands-free control thanks to the firms Gemini chatbot. (The Verge)+ It also revealed a new Samsung-build headset called Project Moohan. (WP $)+ Googles hoping to learn from mistakes it made with Google Glass a decade ago. (Wired $)+ Its new Project Astra could be generative AIs killer app. (MIT Technology Review) 2 The US and UK are on a AI regulation collision course Donald Trumps approach to policing AI is in stark contrast to what the UK is planning. (FT $)+ The new US FTC chair favors a light regulatory touch. (Reuters)+ Hows AI self-regulation going? (MIT Technology Review)3 We dont quite know whats causing a global temperature spike But scientists agree that we should be worried. (New Yorker $)+ The average global temperature could drop slightly next year, though. (New Scientist $)+ Whos to blame for climate change? Its surprisingly complicated. (MIT Technology Review)4 Trumps administration is filling up with tech insiders More venture capitalists and officials are likely to join their ranks. (The Information $)+ These crypto kingpins will be keeping a close eye on proceedings. (FT $)5 What happened after West Virginia revoked access to obesity drugsTeachers and state workers struggled after a pilot drugs program was deemed too expensive. (The Atlantic $) + Weight-loss injections have taken over the internet. But what does this mean for people IRL? (MIT Technology Review)6 Would you buy a car from Amazon?The e-retail giant wants you to sidestep the dealership and purchase from it directly. (Wired $) + While its limited to Hyundai models, other manufacturers will follow. (Forbes $)7 Silicon Valleys perks culture is largely dead No more free massages or artisanal chocolate, sob. (NYT $)8 AI is teaching us more about the Berlin Walls murals From the kinds of paint used, to application techniques. (Ars Technica)9 For $69, you can invest in a rare stegosaurus skeleton The rare fossil is a pretty extreme example of an alternative investment. (Fast Company $)+ New Yorkers can swing by the American Museum of Natural History to see it. (AP News) 10 This New Jersey politician faked his Spotify Wrapped To hide his childrens results and make him appear a bigger Bruce Springsteen fan. (Billboard $)+ What would The Boss himself make of the controversy? (WP $)Quote of the day It could be far worse than any challenge weve previously encountered and far beyond our capacity to mitigate. Jack Szostak, a professor in the University of Chicagos chemistry department, tells the Financial Times about the unprecedented danger posed by synthetic bacteria. The big story A brief, weird history of brainwashing April 2024 On a spring day in 1959, war correspondent Edward Hunter testified before a US Senate subcommittee investigating the effect of Red China Communes on the United States. Hunter introduced them to a supposedly scientific system for changing peoples minds, even making them love things they once hated. Much of it was baseless, but Hunters sensational tales still became an important part of the disinformation that fueled a mind-control race, with the US government pumping millions of dollars into research on brain manipulation during the Cold War. But while the science never exactly panned out, residual beliefs fostered by this bizarre conflict continue to play a role in ideological and scientific debates to this day. Read the full story. Annalee Newitz We can still have nice things A place for comfort, fun and distraction to brighten up your day. (Got any ideas? Drop me a line or tweet 'em at me.) + Deep down in the depths of the Atacama Trench, a new crustacean has been discovered.+ Living in this picturesque Antarctic settlement comes with a catchyou have to have your appendix removed before you can move in.+ Just when you thought sweet potato couldnt get any better, it turns out it makes pretty tasty macaroons.+ If youre looking to introduce kids to the joy of sci-fi, these movies are a great place to start.

Its no secret that the current AI boom is using up immense amounts of energy. Now we have a better idea of how much. A new paper, from a team at the Harvard T.H. Chan School of Public Health, examined 2,132 data centers operating in the United States (78% of all facilities in the country). These facilitiesessentially buildings filled to the brim with rows of serversare where AI models get trained, and they also get pinged every time we send a request through models like ChatGPT. They require huge amounts of energy both to power the servers and to keep them cool. Since 2018, carbon emissions from data centers in the US have tripled. For the 12 months ending August 2024, data centers were responsible for 105 million metric tons of CO2, accounting for 2.18% of national emissions (for comparison, domestic commercial airlines are responsible for about 131 million metric tons). About 4.59% of all the energy used in the US goes toward data centers, a figure thats doubled since 2018. Its difficult to put a number on how much AI in particular, which has been booming since ChatGPT launched in November 2022, is responsible for this surge. Thats because data centers process lots of different types of datain addition to training or pinging AI models, they do everything from hosting websites to storing your photos in the cloud. However, the researchers say, AIs share is certainly growing rapidly as nearly every segment of the economy attempts to adopt the technology. Its a pretty big surge, says Eric Gimon, a senior fellow at the think tank Energy Innovation, who was not involved in the research. Theres a lot of breathless analysis about how quickly this exponential growth could go. But its still early days for the business in terms of figuring out efficiencies, or different kinds of chips. Notably, the sources for all this power are particularly dirty. Since so many data centers are located in coal-producing regions, like Virginia, the carbon intensity of the energy they use is 48% higher than the national average. The paper, which was published on arXiv and has not yet been peer-reviewed, found that 95% of data centers in the US are built in places with sources of electricity that are dirtier than the national average. There are causes other than simply being located in coal country, says Falco Bargagli-Stoffi, an author of the paper. Dirtier energy is available throughout the entire day, he says, and plenty of data centers require that to maintain peak operation 24-7. Renewable energy, like wind or solar, might not be as available. Political or tax incentives, and local pushback, can also affect where data centers get built. One key shift in AI right now means that the fields emissions are soon likely to skyrocket. AI models are rapidly moving from fairly simple text generators like ChatGPT toward highly complex image, video, and music generators. Until now, many of these multimodal models have been stuck in the research phase, but thats changing. OpenAI released its video generation model Sora to the public on December 9, and its website has been so flooded with traffic from people eager to test it out that it is still not functioning properly. Competing models, like Veo from Google and Movie Gen from Meta, have still not been released publicly, but if those companies follow OpenAIs lead as they have in the past, they might be soon. Music generation models from Suno and Udio are growing (despite lawsuits), and Nvidia released its own audio generator last month. Google is working on its Astra project, which will be a video-AI companion that can converse with you about your surroundings in real time. As we scale up to images and video, the data sizes increase exponentially, says Gianluca Guidi, a PhD student in artificial intelligence at University of Pisa and IMT Lucca, who is the papers lead author. Combine that with wider adoption, he says, and emissions will soon jump. One of the goals of the researchers was to build a more reliable way to get snapshots of just how much energy data centers are using. Thats been a more complicated task than you might expect, given that the data is dispersed across a number of sources and agencies. Theyve now built a portal that shows data center emissions across the country. The long-term goal of the data pipeline is to inform future regulatory efforts to curb emissions from data centers, which are predicted to grow enormously in the coming years. Theres going to be increased pressure, between the environmental and sustainability-conscious community and Big Tech, says Francesca Dominici, director of the Harvard Data Science Initiative and another coauthor. But my prediction is that there is not going to be regulation. Not in the next four years.

This is today's edition ofThe Download,our weekday newsletter that provides a daily dose of what's going on in the world of technology. Googles new Project Astra could be generative AIs killer app Google DeepMind has announced an impressive grab bag of new products and prototypes that may just let it seize back its lead in the race to turn generative artificial intelligence into a mass-market concern. Top billing goes to Gemini 2.0the latest iteration of Google DeepMinds family of multimodal large language models, now redesigned around the ability to control agentsand a new version of Project Astra, the experimental everything app that the company teased at Google I/O in May. The margins between top-end models like Gemini 2.0 and those from rival labs like OpenAI and Anthropic are now slim. These days, advances in large language models are less about how good they are and more about what you can do with them. And thats where agents come in. MIT Technology Review got to try out Astra in a closed-door live demo last week. It gave us a hint at whats to come. Find out more in the full story. Will Douglas Heaven China banned exports of a few rare minerals to the US. Things could get messier. Casey Crownhart Ive thought more about gallium and germanium over the last week than I ever have before (and probably more than anyone ever should). China banned the export of those materials to the US last week and placed restrictions on others. The move is just the latest drama in escalating trade tensions between the two countries. While the new export bans could have significant economic consequences, this might be only the beginning. China is a powerhouse, and not just in those niche materialsits also a juggernaut in clean energy, and particularly in battery supply chains. So what comes next could have significant consequences for EVs and climate action more broadly. Read the full story.This story is from The Spark, our weekly climate and energy newsletter. Sign up to receive it in your inbox every Wednesday. The must-reads Ive combed the internet to find you todays most fun/important/scary/fascinating stories about technology. 1 Its looking pretty likely 2024 will be the hottest year on recordBut average temperatures are just one way of assessing our warming world. (New Scientist $) + The first few months of 2025 are likely to be hotter than average, too. (Reuters)+ The US is about to make a sharp turn on climate policy. (MIT Technology Review)2 Meta has donated $1 million to Trumps inaugural fund In an effort to strengthen their previously fractious relationship. (WSJ $)+ Mark Zuckerberg isnt the only tech figure seeking the President-elects ear. (Insider $) 3 How China secretly repatriates Uyghurs Even the United Nations is seemingly powerless to stop it. (WP $)+ Uyghurs outside China are traumatized. Now theyre starting to talk about it. (MIT Technology Review)4 How Big Tech decides when to scrub a users digital footprint Murder suspect Luigi Mangiones Instagram has been taken downbut his Goodreads hasnt. (NYT $)+ Why its dangerous to treat public online accounts as the full story. (NY Mag $)5 Russia-backed hackers targeted Ukraines military using criminal toolsWhich makes it even harder to work out who did it. (TechCrunch) 6 What Cruises exit means for the rest of the robotaxi industryAutomakers are becoming frustrated waiting for the technology to mature. (The Verge) + Cruise will focus on developing fully autonomous personal vehicles instead. (NYT $)7 Researching risky pathogens is extremely high stakes The potential for abuse has some researchers worried we shouldnt undertake it at all. (Undark Magazine)+ Meet the scientist at the center of the covid lab leak controversy. (MIT Technology Review)8 Altermagnetism could be computings next big thingIt would lead to faster, more reliable electronic devices. (FT $) 9 Why some people need so little sleep Gene mutations appear to hold at least some of the answers. (Knowable Magazine)+ Babies spend most of their time asleep. New technologies are beginning to reveal why. (MIT Technology Review)10 Inside the creeping normalization of AI movies The worlds largest TV manufacturer wants to make films for people too lazy to change the channel. (404 Media)+ Unsurprisingly, itll push targeted ads, too. (Ars Technica)+ How AI-generated video is changing film. (MIT Technology Review) Quote of the day "They've made him a martyr for all the troubles people have had with their own insurance companies." Felipe Rodriguez, an adjunct professor at the John Jay College of Criminal Justice in New York, explains why murder suspect Luigi Mangione is being lionized online to Reuters. The big story Why AI could eat quantum computings lunch November 2024 Tech companies have been funneling billions of dollars into quantum computers for years. The hope is that theyll be a game changer for fields as diverse as finance, drug discovery, and logistics. But while the field struggles with the realities of tricky quantum hardware, another challenger is making headway in some of these most promising use cases. AI is now being applied to fundamental physics, chemistry, and materials science in a way that suggests quantum computings purported home turf might not be so safe after all. Read the full story. Edd Gent We can still have nice things A place for comfort, fun and distraction to brighten up your day. (Got any ideas? Drop me a line or tweet 'em at me.) + Working life getting you down? These pictures of bygone office malaise will make you feel a whole lot better (or worsethanks Will!) + Gen Z are getting really into documenting their lives via digital cameras, apparently. + If you believe that Alan MacMasters invented the first electric bread toaster, Im sorry to inform you that youve fallen for an elaborate online hoax.+ The case for a better Turing test for AI-generated art.

The Intel 4004, the first commercial microprocessor, was released in 1971. With 2,300 transistors packed into 12mm2, it heralded a revolution in computing. A little over 50 years later, Apples M2 Ultra contains 134 billion transistors. The scale of progress is difficult to comprehend, but the evolution of semiconductors, driven for decades by Moores Law, has paved a path from the emergence of personal computing and the internet to todays AI revolution. But this pace of innovation is not guaranteed, and the next frontier of technological advancesfrom the future of AI to new computing paradigmswill only happen if we think differently. Atomic challenges The modern microchip stretches both the limits of physics and credulity. Such is the atomic precision, that a few atoms can decide the function of an entire chip. This marvel of engineering is the result of over 50 years of exponential scaling creating faster, smaller transistors. But we are reaching the physical limits of how small we can go, costs are increasing exponentially with complexity, and efficient power consumption is becoming increasingly difficult. In parallel, AI is demanding ever-more computing power. Data from Epoch AI indicates the amount of computing needed to develop AI is quickly outstripping Moores Law, doubling every six months in the deep learning era since 2010. These interlinked trends present challenges not just for the industry, but society as a whole. Without new semiconductor innovation, todays AI models and research will be starved of computational resources and struggle to scale and evolve. Key sectors like AI, autonomous vehicles, and advanced robotics will hit bottlenecks, and energy use from high-performance computing and AI will continue to soar. Materials intelligence At this inflection point, a complex, global ecosystemfrom foundries and designers to highly specialized equipment manufacturers and materials solutions providers like Merckis working together more closely than ever before to find the answers. All have a role to play, and the role of materials extends far, far beyond the silicon that makes up the wafer. Instead, materials intelligence is present in almost every stage of the chip production processwhether in chemical reactions to carve circuits at molecular scale (etching) or adding incredibly thin layers to a wafer (deposition) with atomic precision: a human hair is 25,000 times thicker than layers in leading edge nodes. Yes, materials provide a chips physical foundation and the substance of more powerful and compact components. But they are also integral to the advanced fabrication methods and novel chip designs that underpin the industrys rapid progress in recent decades. For this reason, materials science is taking on a heightened importance as we grapple with the limits of miniaturization. Advanced materials are needed more than ever for the industry to unlock the new designs and technologies capable of increasing chip efficiency, speed, and power. We are seeing novel chip architectures that embrace the third dimension and stack layers to optimize surface area usage while lowering energy consumption. The industry is harnessing advanced packaging techniques, where separate chiplets are fused with varying functions into a more efficient, powerful single chip. This is called heterogeneous integration. Materials are also allowing the industry to look beyond traditional compositions. Photonic chips, for example, harness light rather than electricity to transmit data. In all cases, our partners rely on us to discover materials never previously used in chips and guide their use at the atomic level. This, in turn, is fostering the necessary conditions for AI to flourish in the immediate future. New frontiers The next big leap will involve thinking differently. The future of technological progress will be defined by our ability to look beyond traditional computing. Answers to mounting concerns over energy efficiency, costs, and scalability will be found in ambitious new approaches inspired by biological processes or grounded in the principles of quantum mechanics. While still in its infancy, quantum computing promises processing power and efficiencies well beyond the capabilities of classical computers. Even if practical, scalable quantum systems remain a long way off, their development is dependent on the discovery and application of state-of-the-art materials. Similarly, emerging paradigms like neuromorphic computing, modelled on the human brain with architectures mimicking our own neural networks, could provide the firepower and energy-efficiency to unlock the next phase of AI development. Composed of a deeply complex web of artificial synapses and neurons, these chips would avoid traditional scalability roadblocks and the limitations of todays Von Neumann computers that separate memory and processing. Our biology consists of super complex, intertwined systems that have evolved by natural selection, but it can be inefficient; the human brain is capable of extraordinary feats of computational power, but it also requires sleep and careful upkeep. The most exciting step will be using advanced computeAI and quantumto finally understand and design systems inspired by biology. This combination will drive the power and ubiquity of next-generation computing and associated advances to human well-being. Until then, the insatiable demand for more computing power to drive AIs development poses difficult questions for an industry grappling with the fading of Moores Law and the constraints of physics. The race is on to produce more powerful, more efficient, and faster chips to progress AIs transformative potential in every area of our lives. Materials are playing a hidden, but increasingly crucial role in keeping pace, producing next-generation semiconductors and enabling the new computing paradigms that will deliver tomorrows technology. But materials sciences most important role is yet to come. Its true potential will be to take usand AIbeyond silicon into new frontiers and the realms of science fiction by harnessing the building blocks of biology. This content was produced by EMD Electronics. It was not written by MIT Technology Reviews editorial staff.