You can trust VideoGamer. Our team of gaming experts spend hours testing and reviewing the latest games, to ensure you're reading the most comprehensive guide possible. Rest assured, all imagery and advice is unique and original. Check out how we test and review games here The Nintendo Switch 2 is a beast of a handheld that can support Nvidia DLSS, ray-tracing and other modern rendering features. While the console’s handheld display supports VRR (variable refresh rate) technology to smooth out inconsistent framerates, the console’s Docked Mode output does not. Just weeks ago, Nintendo removed mentions of VRR support in Docked Mode from its official website for the upcoming console. However, Digital Foundry has now confirmed that developers have been informed of the feature’s absence when connecting the console to a TV. Nintendo Switch 2 loses VRR in Docked Mode In a recent video, Digital Foundry lead Richard Leadbetter confirmed the curious VRR decision in a deep breakdown of the Nintendo Switch 2 components. Backed by a source of developer documentation, it’s stated that the Dock simply doesn’t support VRR. “As far as Switch 2 developers are concerned, VRR, right now at least, is indeed a function of the internal display only and that there is no support at all right now for VRR over HDMI,” Leadbetter explained.” The respected journalist and video games analyst explained that the dock included with the Nintendo Switch 2 might not support the technology due to its older HDMI port. “The best theory that we have for this is that the dock’s Display Port to HDMI connector doesn’t support standard HDMI VRR, but whatever is the cause, we really hope to see Nintendo provide some sort of solution in due course.” With the constrained specifications of the Nintendo Switch 2, VRR support is a fantastic option. As 120Hz screens become more common, 40fps modes are offering great experiences for titles that can’t hit 60fps. For the Switch 2, this type of mode will be possible in Portable Mode, but not in Docked Mode. If this is indeed a hardware issue, the only way to actually bring VRR to Docked Mode would be with a revised dock sometime in the future which would split the user base. Would Nintendo be comfortable doing this? Well, judging by the company’s past, probably not. Subscribe to our newsletters! By subscribing, you agree to our Privacy Policy and may receive occasional deal communications; you can unsubscribe anytime. Share

This is today's edition of The Download, our weekday newsletter that provides a daily dose of what's going on in the world of technology. Why climate researchers are taking the temperature of mountain snow The Sierra’s frozen reservoir provides about a third of California’s water and most of what comes out of the faucets, shower heads, and sprinklers in the towns and cities of northwestern Nevada. As it melts through the spring and summer, dam operators, water agencies, and communities have to manage the flow of billions of gallons of runoff, storing up enough to get through the dry summer months without allowing reservoirs and canals to flood. The need for better snowpack temperature data has become increasingly critical for predicting when the water will flow down the mountains, as climate change fuels hotter weather, melts snow faster, and drives rapid swings between very wet and very dry periods. In the past, it was hard work to gather this data. Now, a new generation of tools, techniques, and models promises to ease that process, improve water forecasts, and help California and other states manage in the face of increasingly severe droughts and flooding. However, observers fear that any such advances could be undercut by the Trump administration’s cutbacks across federal agencies.Read the full story. —James Temple MIT Technology Review Narrated: What’s next for our privacy? The US still has no federal privacy law. But recent enforcement actions against data brokers may offer some new protections for Americans’ personal information. This is our latest story to be turned into a MIT Technology Review Narrated podcast, which we’re 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 it’s released. The must-reads I’ve combed the internet to find you today’s most fun/important/scary/fascinating stories about technology. 1 The US is warning other countries not to use Huawei’s chips If they do, they may face criminal penalties for breaching US export controls. (FT $)+ The Trump administration has axed the ‘AI Diffusion Rule’ for chips. (WSJ $)+ It may move towards negotiating deals with countries directly. (Bloomberg $) 2 US tech firms are inking AI deals with the Middle EastAmong the biggest of which is Nvidia. (The Guardian) + Tech leaders accompanied Trump on his trip. (WP $)3 A new treatment for inherited breast cancer was trialed successfully The drug olaparib can help to significantly improve survival rates. (BBC)4 TikTok workers fear a new messaging feature could be exploited But the company is pressing ahead with it anyway. (The Information $)5 Apple is working on brain-computer interfaces for its products People with brain implants could one day use them to control their devices. (WSJ $)+ Brain-computer interfaces face a critical test. (MIT Technology Review) 6 What’s next for NASA? The agency is poised for its most radical shakeup in decades. (Ars Technica)+ NASA has made an air traffic control system for drones. (MIT Technology Review)7 Finland is harvesting heat from its data centersBloomberg $) + The next data center hub? India. (FT $)+ These four charts sum up the state of AI and energy. (MIT Technology Review)8 Airbnb wants to become the next everything-appWired $) + Hotel-like services, anyone? (NYT $)+ Its host features have been overhauled, too. (The Verge)9 The FBI is buying new tech to help it see through wallsThanks to radar. (New Scientist $) 10 Baidu is planning to launch its robotaxis in Europe 🚗 In a bid to extend its competitive advantage overseas. (WSJ $)+ How Wayve’s driverless cars will meet one of their biggest challenges yet. (MIT Technology Review) Quote of the day “It's literally Einstein's proverbial definition of insanity: doing the same thing over and over again and expecting a different result.” —Entrepreneur Arnaud Bertrand reflects on America’s latest attempt to rein in Huawei in a post on X. One more thing How DeepSeek ripped up the AI playbook—and why everyone’s going to follow its lead When the Chinese firm DeepSeek dropped a large language model called R1 at the start of this year, it sent shock waves through the US tech industry. Not only did R1 match the best of the homegrown competition, it was built for a fraction of the cost—and given away for free. DeepSeek has now suddenly become the company to beat. What exactly did it do to rattle the tech world so fully? Is the hype justified? And what can we learn from the buzz about what’s coming next? Here’s what you need to know.   —Will Douglas Heaven 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.)+ Happy birthday to the one, the only Stevie Wonder—75 years young this week.+ A scooped bagel? Not on my watch. 🥯+ A few tips on how to navigate some of life’s trickier conversations with ease.+ Everyone’s got a random junk drawer. Here’s how to get it under control.

TNW Backstage is going Double Dutch this week — in more ways than one. In the second episode of our new podcast, which goes behind the scenes of TNW Conference and the tech shaping our world, we’re joined by two special guests. Both of them are based in — you guessed it — the Netherlands. Together, they brew a distinctively orange blend of tech and humour, which has become a TNW trademark. One of them is content creator Derek Mitchell, the mind behind comedy brand Let’s Double Dutch. Mitchell’s viral videos exploring culture and identity in the Netherlands have earned global acclaim. His wildly popular sketches have now racked up over 250 million views. In TNW Backstage, we revisit his appearance at last year’s conference. During the talk, Mitchell reveals how he found his voice online — and how you can too.The 💜 of EU techThe latest rumblings from the EU tech scene, a story from our wise ol' founder Boris, and some questionable AI art. It's free, every week, in your inbox. Sign up now! Now, you may be wondering why a comedian was appearing at a tech conference. Well, there’s no one better to explain than this week’s second guest: Boris, the co-founder of TNW. Boris drops by TNW Towers to explain how humour fuels startup ideas — and lightens the load of tech work. He also looks back at memorable moments from TNW Conference. Among his favourites are a risky rib at Amazon’s CTO, wakeboarding through Amsterdam canals, and, um, cow poop bingo. (It’s another Dutch thing, apparently.) Our terrifically talented head of content, Oliviana Bailey, then gives a sneak peek at the fun and games at this year’s events. It’s a distinct break from the debut episode of the podcast, which explored the evolving data security landscape — and Meta’s controversial “pay or consent” model — with Ron de Jesus, the world’s first Field Chief Privacy Officer. But the contrast encapsulates the two sides of TNW Conference: serious business and festival vibes. You can check out both episodes of the podcast here. To celebrate its launch, we’re also offering an exclusive discount on tickets for TNW Conference, which takes place on June 19 and 20. You’ll find the offer hidden in each episode of TNW Backstage. Thanks for reading — and now, for listening too. Story by Thomas Macaulay Managing editor Thomas is the managing editor of TNW. He leads our coverage of European tech and oversees our talented team of writers. Away from work, he e (show all) Thomas is the managing editor of TNW. He leads our coverage of European tech and oversees our talented team of writers. Away from work, he enjoys playing chess (badly) and the guitar (even worse). Get the TNW newsletter Get the most important tech news in your inbox each week. Also tagged with

Author(s): Towards AI Editorial Team Originally published on Towards AI. Over the past two years, we’ve helped teams design and deploy real-world LLM systems — RAG pipelines, copilots that actually reduce load, PoCs that became products, and cut down hallucinations. One year ago, we decided to put everything we knew about building architectures around LLMs — the stack, the mistakes, the gotchas, the strategies — into a single guide: Building LLMs for Production — “the most comprehensive textbook to date on building LLM applications,” as Jerry Liu (Founder & CEO, LlamaIndex) put it. The response was amazing. People read it. People built with it. People shared it. But a few months in, our DMs started filling up: “Has the book been updated?” “Does it cover the latest models like o3 or Gemini 2.5?” “Can I use [X vector DB] instead?” “How do I choose the right model for my use case?” “What if I want to do [insert new approach] that isn’t in the book?” Fair. The landscape’s shifting fast. Inference got scaled. SLMs showed up. Context windows stretched. Costs dropped. Everything moved. If AI has taught us anything, it’s to think AI-first — not just to keep up, but to build in ways that scale. So instead of answering each DM (which we still do), we took a step back. And we decided to build something that answers all of it, now and as things evolve. The result? From Beginner to Advanced LLM Developer A 60+ hour, hands-on course that takes you from “I can prompt ChatGPT” → to deploying a production-grade RAG system with a real front end. But we didn’t just pack it with knowledge — we designed it to evolve with the field. Here’s what you walk away with: ✅ A repeatable pipeline that adapts with tools, not trends ✅ A deep instinct for how to think like an AI engineer ✅ Lifetime access and weekly updates as the ecosystem changes ✅ A private Slack for graduates + a 70,000+ builder community on Discord Because staying current isn’t enough — you also need confidence that what you ship today still holds tomorrow. That’s why we’re now running monthly live cohorts — so you stay sharp, supported, and up to date. The next cohort kicks off June 1 with a live welcome call with our CEO. Launch price: $249 (75 % off) — zero risk thanks to a 30-day money-back guarantee. 👉 Join the course here The results speak for themselves: “The course greatly expanded my knowledge of building and assessing RAG pipelines.” — Eoin McGrath “Best course out there to become an AI engineer. Planning to build my own startup based on the learnings.” — Abhijit L. “From zero to hero as an LLM Developer — a clear path to build LLM applications that can change your career.” — Luca Tanieli Even industry voices you know have shared support: “A resource I’ll return to again and again, no matter how fast the AI landscape shifts.” — Tina Huang, Lonely Octopus This course is for you: You know Python but haven’t taken an LLM past the notebook. If you’re frustrated by shallow tutorials and fragmented docs… If you want to build things that work, not just read about them… If you’re ready to take LLMs seriously and want a proven structure… There’s a roadmap. And it’s working. The next cohort starts June 1st. As soon as you join, you get full access to all course material — no need to wait for the live kickoff. You can start building right away. What You’ll Learn: 🧠 LLM Basics & Prompt Mastery Transformers, tokenization, and prompting that actually reduces hallucinations 🔍 Retrieval-Augmented Generation (RAG) Chunking, embedding models, re-ranking, query rewriting, eval, and feedback loops 🎨 Fine-Tuning LoRA, adapters, and domain-specific models that actually perform 🤖 Tool & API Integration Function calling, external tools, and chained agent workflows 🚀 Deployment & Cost Control Gradio, Streamlit, latency fixes, caching, logging, monitoring, cost tracking 🏆 Capstone Project & Certification Build and ship your own LLM app — get feedback, and leave with a portfolio-ready build If you’re thinking, “This sounds great, but what if it’s not for me?” — we get it. That’s why the course comes with a 30-day, no-questions-asked money-back guarantee. Try it. Dive into the material. If it doesn’t meet your expectations, we’ll refund you in full. 👉 Secure your spot for the June 1st cohort Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor. Published via Towards AI

AI company Schemata has come out of stealth and announced $5 million in seed funding to accelerate the development of its training platform. The company’s 3D reality capture, spatial AI, and contextual information can help the defense and enterprise sectors more rapidly and cost-effectively create photorealistic and interactive training programs. The new platform is already being piloted by the Department of Defense. Schemata’s technology converts 2D images into photorealistic 3D environments using 3D reality capture. It then uses spatial AI to apply contextual information to those assets, as well as an AI Instructor so that users can engage with the 3D simulations through voice or text. The result is personalized AI-enabled training assistance that can be created 10 times faster and up to 85% more cost-effectively than traditional content creation means, the company said. The startup’s origins trace back to Stanford University, where founders James Brown and Huy Nguyen first developed the concept for an AI-driven simulation platform leveraging Brown’s insights on military training and simulation and Nguyen’s expertise in 3D graphics and computer vision​. The pair brought on former congressional staffer and co-founder Quin Roberts, and left Stanford last June to build Schemata. “By harnessing the latest advances in generative AI’s reasoning capabilities, we can more effectively parse complex 3D data—and, in turn, gain deeper insights into real-world environments,” said Brown, Schemata CEO, i an statement. “In the Marine Corps, everything I did was centered around large-scale physical operations on complex terrain and equipment. At Stanford, we saw how AI was transforming knowledge work, and we wanted to bring those same advancements to those who work out in the physical world”. 3D reality capture is the process of using technologies like laser scanners, drones, and photogrammetry to digitally reconstruct physical spaces in 3D​. The result is a highly accurate virtual model of a real environment. Spatial AI refers to AI systems that can understand and interact within these spaces – in other words, machines navigating the environments with human-like perception​. By combining these two technologies, Schemata can quickly generate a lifelike virtual replica of any location and embed an intelligent virtual instructor to guide users through it. Within these digital environments, the AI Instructor can dynamically interact with end-users: answering questions, highlighting objects, and adapting scenarios in real time. For example, an operator using a device loaded with Schemata’s models can ask a question out loud and receive instant guidance. Training modules that once took specialists months to build can now be generated in hours​ directly from a company’s existing training materials. “Our AI Instructor is like having a dedicated expert by your side 24/7 in a physical environment,” said Quin Roberts, Schemata’s COO, in a statement. “It can instantly analyze a 3D scene, point out safety hazards or operational steps, and respond to the user’s voice or text queries in real-time. This level of interactive, on-demand instruction simply isn’t available with conventional tools.” The timing of Schemata’s innovation aligns with a surging demand for virtual training across defense and industry. In the defense sector, the U.S. military already invests over $12 billion per year in simulation-based training​. Yet, these investments have primarily focused on high-end systems for pilots and specialized units. Schemata’s platform aims to broaden that reach. The platform is already being piloted in sectors like defense, energy, and infrastructure, where training often involves complex environments and high stakes. Demand is strong, with Schemata already being awarded $3.4 million in Department of Defense (DoD) contracts over the past eight months. Schemata’s oversubscribed funding round includes investments from prominent venture capital firms, led by Owl Ventures, with participation from a16z speedrun, Alumni Ventures, Anorak Ventures, and Time Zero Capital. Last summer, Schemata was selected for the third cohort of a16z’s speedrun startup program, with an acceptance rate of less than 1%. “We see a future where spatially-aware AI agents operate with a full context of the physical world—just like frontier language models now operate across digital systems,” said Nguyen, Schemata CTO, in a statement. “That’s why we’re connecting AI to rich 3D environments with embedded semantic data. Virtual training is just the beginning of what it can unlock”. GB Daily Stay in the know! Get the latest news in your inbox daily Read our Privacy Policy Thanks for subscribing. Check out more VB newsletters here. An error occured.

This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. Why climate researchers are taking the temperature of mountain snow The Sierra’s frozen reservoir provides about a third of California’s water and most of what comes out of the faucets, shower heads, and sprinklers in the towns and cities of northwestern Nevada. As it melts through the spring and summer, dam operators, water agencies, and communities have to manage the flow of billions of gallons of runoff, storing up enough to get through the dry summer months without allowing reservoirs and canals to flood. The need for better snowpack temperature data has become increasingly critical for predicting when the water will flow down the mountains, as climate change fuels hotter weather, melts snow faster, and drives rapid swings between very wet and very dry periods. In the past, it was hard work to gather this data. Now, a new generation of tools, techniques, and models promises to ease that process, improve water forecasts, and help California and other states manage in the face of increasingly severe droughts and flooding. However, observers fear that any such advances could be undercut by the Trump administration’s cutbacks across federal agencies. Read the full story. —James Temple MIT Technology Review Narrated: What’s next for our privacy? The US still has no federal privacy law. But recent enforcement actions against data brokers may offer some new protections for Americans’ personal information. This is our latest story to be turned into a MIT Technology Review Narrated podcast, which we’re 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 it’s released. The must-reads I’ve combed the internet to find you today’s most fun/important/scary/fascinating stories about technology. 1 The US is warning other countries not to use Huawei’s chipsIf they do, they may face criminal penalties for breaching US export controls. (FT $)+ The Trump administration has axed the ‘AI Diffusion Rule’ for chips. (WSJ $)+ It may move towards negotiating deals with countries directly. (Bloomberg $) 2 US tech firms are inking AI deals with the Middle EastAmong the biggest of which is Nvidia. (The Guardian)+ Tech leaders accompanied Trump on his trip. (WP $) 3 A new treatment for inherited breast cancer was trialed successfullyThe drug olaparib can help to significantly improve survival rates. (BBC) 4 TikTok workers fear a new messaging feature could be exploitedBut the company is pressing ahead with it anyway. (The Information $) 5 Apple is working on brain-computer interfaces for its productsPeople with brain implants could one day use them to control their devices. (WSJ $)+ Brain-computer interfaces face a critical test. (MIT Technology Review) 6 What’s next for NASA?The agency is poised for its most radical shakeup in decades. (Ars Technica)+ NASA has made an air traffic control system for drones. (MIT Technology Review) 7 Finland is harvesting heat from its data centersWhile no data center is good for the environment, this helps lessen their footprint. (Bloomberg $)+ The next data center hub? India. (FT $)+ These four charts sum up the state of AI and energy. (MIT Technology Review) 8 Airbnb wants to become the next everything-appIt wants to expand beyond vacations and into a community platform, apparently. (Wired $)+ Hotel-like services, anyone? (NYT $)+ Its host features have been overhauled, too. (The Verge) 9 The FBI is buying new tech to help it see through wallsThanks to radar. (New Scientist $) 10 Baidu is planning to launch its robotaxis in Europe In a bid to extend its competitive advantage overseas. (WSJ $)+ How Wayve’s driverless cars will meet one of their biggest challenges yet. (MIT Technology Review) Quote of the day “It’s literally Einstein’s proverbial definition of insanity: doing the same thing over and over again and expecting a different result.” —Entrepreneur Arnaud Bertrand reflects on America’s latest attempt to rein in Huawei in a post on X. One more thing How DeepSeek ripped up the AI playbook—and why everyone’s going to follow its lead When the Chinese firm DeepSeek dropped a large language model called R1 at the start of this year, it sent shock waves through the US tech industry. Not only did R1 match the best of the homegrown competition, it was built for a fraction of the cost—and given away for free. DeepSeek has now suddenly become the company to beat. What exactly did it do to rattle the tech world so fully? Is the hype justified? And what can we learn from the buzz about what’s coming next? Here’s what you need to know.   —Will Douglas Heaven 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.)+ Happy birthday to the one, the only Stevie Wonder—75 years young this week.+ A scooped bagel? Not on my watch. + A few tips on how to navigate some of life’s trickier conversations with ease.+ Everyone’s got a random junk drawer. Here’s how to get it under control.

TNW Backstage is going Double Dutch this week — in more ways than one. In the second episode of our new podcast, which goes behind the scenes of TNW Conference and the tech shaping our world, we’re joined by two special guests. Both of them are based in — you guessed it — the Netherlands. Together, they brew a distinctively orange blend of tech and humour, which has become a TNW trademark. One of them is content creator Derek Mitchell, the mind behind comedy brand Let’s Double Dutch. Mitchell’s viral videos exploring culture and identity in the Netherlands have earned global acclaim. His wildly popular sketches have now racked up over 250 million views. The 💜 of EU techThe latest rumblings from the EU tech scene, a story from our wise ol' founder Boris, and some questionable AI art. It's free, every week, in your inbox. Sign up now!In TNW Backstage, we revisit his appearance at last year’s conference. During the talk, Mitchell reveals how he found his voice online — and how you can too. Now, you may be wondering why a comedian was appearing at a tech conference. Well, there’s no one better to explain than this week’s second guest: Boris, the co-founder of TNW. Boris drops by TNW Towers to explain how humour fuels startup ideas — and lightens the load of tech work. He also looks back at memorable moments from TNW Conference. Among his favourites are a risky rib at Amazon’s CTO, wakeboarding through Amsterdam canals, and, um, cow poop bingo. (It’s another Dutch thing, apparently.) Our terrifically talented head of content, Oliviana Bailey, then gives a sneak peek at the fun and games at this year’s events. It’s a distinct break from the debut episode of the podcast, which explored the evolving data security landscape — and Meta’s controversial “pay or consent” model — with Ron de Jesus, the world’s first Field Chief Privacy Officer. But the contrast encapsulates the two sides of TNW Conference: serious business and festival vibes. You can check out both episodes of the podcast here. To celebrate its launch, we’re also offering an exclusive discount on tickets for TNW Conference, which takes place on June 19 and 20. You’ll find the offer hidden in each episode of TNW Backstage. Thanks for reading — and now, for listening too. Story by Thomas Macaulay Managing editor Thomas is the managing editor of TNW. He leads our coverage of European tech and oversees our talented team of writers. Away from work, he e (show all) Thomas is the managing editor of TNW. He leads our coverage of European tech and oversees our talented team of writers. Away from work, he enjoys playing chess (badly) and the guitar (even worse). Get the TNW newsletter Get the most important tech news in your inbox each week. Also tagged with

Credit: Hisense Deal pricing and availability subject to change after time of publication. Learn more about how we select deals. SAVE $300: The Hisense C2 TriChroma Laser portable projector is on sale for $1,699.99 at Amazon as of May 14, which is a 15% discount and drops it down to its best price ever. Opens in a new window Credit: Hisense Hisense C2 TriChroma Laser portable projector $1,699.99 at Amazon $1,999.99 Save $300 Projectors have come a long way in recent years — to the point where some rival big screen 4K TVs. The Hisense C2 TriChroma Laser series, released last fall, is one of those rivals. Of course, that groundbreaking cinematic experience doesn't come cheap. The C2 TriChroma Laser series includes three models, with the most affordable of the trio costing $1,999.99. As of May 14, however, the standard Hisense C2 portable projector is on sale for $1,699.99. That's 15% off and its lowest price on record. Sure, that's still pretty steep, but it's $300 less than you'd spend at full retail price.All three models — including the standard C2 projector — are equipped with Hisense's Tri-Chroma triple color laser, which ensures color accuracy with a massive color spectrum. They also support 3D viewing, adding another dimension to your viewing experience. Projection size ranges from 65 to 300 inches and thanks to 360-degree horizontal rotation and 135-degree vertical adjustment, you'll have unparalleled viewing flexibility. You can watch movies on the ceiling, set up a massive outdoor screen, or stream workout videos on the wall of your home gym. Each model also offers auto keystone correction and auto focus, smart wall color adaptation, and obstacle avoidance, ensuring a top-notch viewing experience instantly.The standard C2 model, as a relatively budget-friendly option, only boasts 2,000 lumens of brightness (versus the 3,000 lumens of the C2 Ultra). That's still plenty bright, though, unless you're in an extremely sunny atmosphere. Just dim the lights and close the curtains and you should be fine. It also lacks some of the higher-end features, like a built-in subwoofer and IMAX Enhanced certification. However, it still offers key technologies like Dolby Vision and HDR10+, as well as JBL 2.0 channel audio. Related Stories Mashable Deals Want more hand-picked deals from our shopping experts? Sign up for the Mashable Deals newsletter. By clicking Sign Me Up, you confirm you are 16+ and agree to our Terms of Use and Privacy Policy. Thanks for signing up! If you prefer a portable and flexible TV experience, it's hard to beat the cinematic capabilities of the Hisense C2 TriChroma Laser projector — particularly when it's on sale. Topics Amazon Gadgets Christina Buff Christina Buff is a Nashville-based freelance writer for who covers shopping with a splash of entertainment. If you’re ever wondering what streaming service you need to watch something (and the cheapest way to sign up for it), she’s your girl.Christina received a B.S. in Business Communication (concentration in writing) from Stevenson University and began her professional journey writing and editing press releases. Since then, she’s written content for a marketing agency, blogged for celebrities, and covered local news, politics, women’s lifestyle, fashion, beauty, and just about everything in between for various publications.When she’s not writing, she’s probably enjoying live music, studying human design, or embroidering and upcycling clothes. You can follow her on Instagram at @touchinfinity.

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More Google DeepMind today pulled the curtain back on AlphaEvolve, an artificial-intelligence agent that can invent brand-new computer algorithms — then put them straight to work inside the company’s vast computing empire. AlphaEvolve pairs Google’s Gemini large language models with an evolutionary approach that tests, refines, and improves algorithms automatically. The system has already been deployed across Google’s data centers, chip designs, and AI training systems — boosting efficiency and solving mathematical problems that have stumped researchers for decades. “AlphaEvolve is a Gemini-powered AI coding agent that is able to make new discoveries in computing and mathematics,” explained Matej Balog, a researcher at Google DeepMind, in an interview with VentureBeat. “It can discover algorithms of remarkable complexity — spanning hundreds of lines of code with sophisticated logical structures that go far beyond simple functions.” The system dramatically expands upon Google’s previous work with FunSearch by evolving entire codebases rather than single functions. It represents a major leap in AI’s ability to develop sophisticated algorithms for both scientific challenges and everyday computing problems. Inside Google’s 0.7% efficiency boost: How AI-crafted algorithms run the company’s data centers AlphaEvolve has been quietly at work inside Google for over a year. The results are already significant. One algorithm it discovered has been powering Borg, Google’s massive cluster management system. This scheduling heuristic recovers an average of 0.7% of Google’s worldwide computing resources continuously — a staggering efficiency gain at Google’s scale. The discovery directly targets “stranded resources” — machines that have run out of one resource type (like memory) while still having others (like CPU) available. AlphaEvolve’s solution is especially valuable because it produces simple, human-readable code that engineers can easily interpret, debug, and deploy. The AI agent hasn’t stopped at data centers. It rewrote part of Google’s hardware design, finding a way to eliminate unnecessary bits in a crucial arithmetic circuit for Tensor Processing Units (TPUs). TPU designers validated the change for correctness, and it’s now headed into an upcoming chip design. Perhaps most impressively, AlphaEvolve improved the very systems that power itself. It optimized a matrix multiplication kernel used to train Gemini models, achieving a 23% speedup for that operation and cutting overall training time by 1%. For AI systems that train on massive computational grids, this efficiency gain translates to substantial energy and resource savings. “We try to identify critical pieces that can be accelerated and have as much impact as possible,” said Alexander Novikov, another DeepMind researcher, in an interview with VentureBeat. “We were able to optimize the practical running time of [a vital kernel] by 23%, which translated into 1% end-to-end savings on the entire Gemini training card.” Breaking Strassen’s 56-year-old matrix multiplication record: AI solves what humans couldn’t AlphaEvolve solves mathematical problems that stumped human experts for decades while advancing existing systems. The system designed a novel gradient-based optimization procedure that discovered multiple new matrix multiplication algorithms. One discovery toppled a mathematical record that had stood for 56 years. “What we found, to our surprise, to be honest, is that AlphaEvolve, despite being a more general technology, obtained even better results than AlphaTensor,” said Balog, referring to DeepMind’s previous specialized matrix multiplication system. “For these four by four matrices, AlphaEvolve found an algorithm that surpasses Strassen’s algorithm from 1969 for the first time in that setting.” The breakthrough allows two 4×4 complex-valued matrices to be multiplied using 48 scalar multiplications instead of 49 — a discovery that had eluded mathematicians since Volker Strassen’s landmark work. According to the research paper, AlphaEvolve “improves the state of the art for 14 matrix multiplication algorithms.” The system’s mathematical reach extends far beyond matrix multiplication. When tested against over 50 open problems in mathematical analysis, geometry, combinatorics, and number theory, AlphaEvolve matched state-of-the-art solutions in about 75% of cases. In approximately 20% of cases, it improved upon the best known solutions. One victory came in the “kissing number problem” — a centuries-old geometric challenge to determine how many non-overlapping unit spheres can simultaneously touch a central sphere. In 11 dimensions, AlphaEvolve found a configuration with 593 spheres, breaking the previous record of 592. How it works: Gemini language models plus evolution create a digital algorithm factory What makes AlphaEvolve different from other AI coding systems is its evolutionary approach. The system deploys both Gemini Flash (for speed) and Gemini Pro (for depth) to propose changes to existing code. These changes get tested by automated evaluators that score each variation. The most successful algorithms then guide the next round of evolution. AlphaEvolve doesn’t just generate code from its training data. It actively explores the solution space, discovers novel approaches, and refines them through an automated evaluation process — creating solutions humans might never have conceived. “One critical idea in our approach is that we focus on problems with clear evaluators. For any proposed solution or piece of code, we can automatically verify its validity and measure its quality,” Novikov explained. “This allows us to establish fast and reliable feedback loops to improve the system.” This approach is particularly valuable because the system can work on any problem with a clear evaluation metric — whether it’s energy efficiency in a data center or the elegance of a mathematical proof. From cloud computing to drug discovery: Where Google’s algorithm-inventing AI goes next While currently deployed within Google’s infrastructure and mathematical research, AlphaEvolve’s potential reaches much further. Google DeepMind envisions applications in material sciences, drug discovery, and other fields requiring complex algorithmic solutions. “The best human-AI collaboration can help solve open scientific challenges and also apply them at Google scale,” said Novikov, highlighting the system’s collaborative potential. Google DeepMind is now developing a user interface with its People + AI Research team and plans to launch an Early Access Program for selected academic researchers. The company is also exploring broader availability. The system’s flexibility marks a significant advantage. Balog noted that “at least previously, when I worked in machine learning research, it wasn’t my experience that you could build a scientific tool and immediately see real-world impact at this scale. This is quite unusual.” As large language models advance, AlphaEvolve’s capabilities will grow alongside them. The system demonstrates an intriguing evolution in AI itself — starting within the digital confines of Google’s servers, optimizing the very hardware and software that gives it life, and now reaching outward to solve problems that have challenged human intellect for decades or centuries. Daily insights on business use cases with VB Daily If you want to impress your boss, VB Daily has you covered. We give you the inside scoop on what companies are doing with generative AI, from regulatory shifts to practical deployments, so you can share insights for maximum ROI. Read our Privacy Policy Thanks for subscribing. Check out more VB newsletters here. An error occured.

Skip to content News Paleontology This exquisite Archaeopteryx fossil reveals how flight took off in birds Analyses of the fossil show never-before-seen feathers and bones from the first known bird When viewed under ultraviolet light, the soft tissues and fine skeletal details of the new Archaeopteryx specimen become visible. Delaney Drummond, ©Field Museum By Carolyn Gramling 1 hour ago An exceptionally preserved specimen of the ancient bird Archaeopteryx offers the most detailed window yet into the evolution of flight, researchers report online May 14 in Nature. The remarkable preservation of the specimen — the 14th Archaeopteryx ever unearthed — means that researchers can study aspects of the ancient bird that were previously difficult to discern, from the anatomy of its skull to the arrangement of its feathers to the soft tissues on its hands and feet. Sign up for our newsletter We summarize the week's scientific breakthroughs every Thursday.