The good room Gustavo Utrabo would like to be in is imagined; its only representation is this watercolour paintingCredit:Gustavo UtraboIt is a long and hot summers afternoon in Brazil. The room is articulated by three elements: the ground, a roof that provides shade, and a constructed nature all around. It is an ever-changing room, where walls change colour, and need to be trimmed and taken care of. In time, it alters our perception of what a room could be. Gustavo UtraboExplore more from the good rooms series here, or in the Good rooms + AR House issue2025-01-06Reuben J BrownShare

At CES 2025, Circular is bringing more finishes, a titanium build, eight days of battery life, and digital sizing to its second-generation smart ring.

The new Roborock Saros Z70 robot vacuum features an arm that grabs toys, socks, and other small obstacles to clean your floors more thoroughly.

These new AI-powered features offer much more than just a boost to picture quality.

Multimodal AI - Use CasesLutz FingerMultimodality is set to redefine how enterprises leverage AI in 2025. Imagine an AI that understands not just text but also images, audio, and other sensor data. Humans are naturally multimodal. However, humans are limited in how much input we can process. Take healthcare as an example, during my time at Google Health, I heard many stories where patients overwhelmed doctors with data:Imagine a patient with atrial fibrillation (AFIB) showing up with five years of detailed sleep data collected from their smartwatch. Or take the cancer patient arriving with a 20-pound stack of medical records documenting every treatment theyve had. Both of these situations are very real. For doctors, the challenge is the same: separating the signal from the noise.Whats needed is an AI that can summarize and highlight the key points. Large language models, like ChatGPT, already do this with text, pulling out the most relevant information. But what if we could teach AI to do the same with other types of data like images, time series, or lab results?How Does Multimodality AI Work?To understand how multimodality works, lets start with the fact that AI needs data both to be trained and to make predictions. Multimodal AI is designed to handle diverse data sources text, images, audio, video, and even time-series data at the same time. By combining these inputs, multimodal AI offers a richer, more comprehensive understanding of the problems it tackles.Multimodal AI is more of a discovery tool. The different data modalities are stored by the AI. Once a new data point is input, the AI finds topics that are close. For example, by inputting the sleep data from someone's smartwatch alongside information about their atrial fibrillation (AFIB) episodes, the doctor might find indications of sleep apnea.Read More: AI Agents In 2025: What Enterprise Leaders Need To KnowNote that this is based on "closeness," not correlation. It is the scaled-up version of what Amazon once popularized: "people who shopped for this item also bought this item." In this case, its more like: "People with this type of sleep pattern have also been diagnosed with AFIB."Multimodal Explained: Encoders, Fusion and DecodersA multimodal AI system consists of three main components: Encoders, Fusion and Decoders.Encoding Any ModalityEncoders convert raw data (e.g., text, images, sound, log files, etc.) into a representation the AI can work with. These are called vectors, which are stored in a latent space. To simplify, think of this process as storing an item in a warehouse (latent space), where each item has a specific location (vector). Encoders can process virtually anything: images, text, sound, videos, log files, IoT (sensor) information, time series you name it.Encoding Multimodal Information into a Latent Vector SpaceLutz FingerFusion Mechanism: Combining ModalitiesWhen working with one type of data, like images, encoding is enough. But with multiple types images, sounds, text, or time-series data we need to fuse the information to find whats most relevant.Decoders: Generating Outputs We UnderstandDecoders decodes the information from the latent space aka the warehouse and deliver it to us. It moves from raw, abstract information to something we understand. For example, finding an image of a "house."Encoding and Decoding Multimodal DataLutz FingerIf you want to learn more about encoding, decoding, and reranking, join my eCornell Online Certificate course on Designing and Building AI Solutions. Its a no-coding program that explores all aspects of AI solutions.Transforming eCommerce with MultimodalityLets look at another example: eCommerce. Amazons interface hasnt changed much in 25 years you type a keyword, scroll through results, and hope to find what you need. Multimodality can transform this experience by letting you describe a product, upload a photo, or provide context to find your perfect match.Fixing Search with Multimodal AIAt r2decide, a company a few Cornellians and I started, were using multimodality to merge Search, Browse, and Chat into one seamless flow. Our customers are eCommerce companies tired of losing revenue because their users couldnt find what they needed. At the core of our solution is multimodal AI.For example, in an online jewelry store, a user searching for green would in the past only see green jewelry if the word green appeared in the product text. Since r2decide's AI also encodes images into a shared latent space (e.g., warehouse), it finds green across all modalities. The items are then re-ranked based on the user's past searches and clicks to ensure they receive the most relevant "green" options.Using r2decide Multimodal Search, users will find what they are looking for.Lutz FingerUsers can also search for broader contexts, like "wedding," "red dress," or "gothic." The AI encodes these inputs into the latent space, matches them with suitable products, and displays the most relevant results. This capability even extends to brand names like Swarovski, surfacing relevant items even if the shop doesnt officially carry Swarovski products.Using r2decide's Multimodal Search, users will even find items that "look" like jewelry from a ... [+] competitor.Lutz FingerAI-Generated Nudges to Give Chat-Like AdviceAlongside search results, R2Decide also generates AI-driven nudges contextual recommendations or prompts designed to enhance the user experience. These nudges are powered by AI agents, as I described in my post on agentic AI yesterday. Their purpose is to guide users effortlessly toward the most relevant options, making the search process intuitive, engaging, and effective.Multimodality in 2025: Infinite Possibilities for EnterprisesMultimodality is transforming industries, from healthcare to eCommerce. And it doesnt stop there. Startups like TC Labs use multimodal AI to streamline engineering workflows, boosting efficiency and quality, while Toyota uses it for interactive, personalized customer assistance.2025 will be the year multimodal AI transforms how enterprises work. Follow me here on Forbes, or on LinkedIn for more of my 2025 AI predictions.

Don Pettit is on a roll. Just days after sharing a breathtaking shot of Earth, the American astronaut has followed up with an equally stunning shot of a waning crescent moon, or, as Pettit put it: a glowing fingernail clipping in the sky.Glowing fingernail clipping in the sky; its the waning crescent moon. pic.twitter.com/zXxkz4k4BZ Don Pettit (@astro_Pettit) January 4, 2025The striking image was captured from the International Space Station (ISS) 250 miles above Earth and also shows the glow of a sunrise or sunset over Earth.Recommended VideosPettit, who at 69 is NASAs oldest serving astronaut, arrived at the orbital outpost in September, and since then, hes been dazzling his social media followers with a steady stream of impressive imagery showing Earth, space, and even a crewed SpaceX spacecraft heading home at high speed.Please enable Javascript to view this contentPettit is one of a number of astronauts who have earned a reputation for producing amazing photographic work while living and working aboard the space station.RelatedThe club also includes NASAs Matthew Dominick, who departed the ISS in October following a six-month stay. During his time in orbit, Dominick took an astonishing half a million images, the best of which he posted on sites like X. Dominick also liked to share details about the camera, lens, and exposures that he used for each shot, giving photography enthusiasts back on terra firma some valuable insight into his work.Another highly talented photographer to have visited the ISS is Thomas Pesquet. During his most recent stay at the space station in 2021, the French astronaut went to great lengths to ensure that he could capture some breathtakingly beautiful images of Earth from orbit.Editors Recommendations

html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN" "http://www.w3.org/TR/REC-html40/loose.dtd" Just about a year ago, a startup named DoublePoint launched a gesture control app that lets smartwatch users control phones, tablets, and headsets, among other devices. The Apple Watch has finally received its own version.In the lead-up to CES 2025, Doublepoint introduced the WowMouse app for Apple Watch, which boasts a few algorithmic refinements and a partnership with Bosch. The apps Android version has already raked in over 100,000 downloads, says the company.Recommended VideosThe premise of WowMouse is rather simple. With an Apple Watch on your wrist, you can control your Mac with cursor and click movements. But thats just the start, as DoublePoint apparently has bigger ambitions.Doublepoint TechnologiesDoublepoint plans to expand connectivity in the near future to include control of any Bluetooth-enabled device, says the company. On top of that, the software will be open-sourced so that developers can build on the existing work. To that end, DoublePoint has worked with Bosch Sensortec to integrate its gesture algorithms with the latters inertial measurement units (IMUs) to allow for energy-efficient and precise gesture sensing.RelatedBosch Sensortec is already a well-known player in the phone and tablet segment that provides MEMS sensors used for a wide range of chores such as activity measurement, gesture detection, and image stabilization.DoublePoints partnership with Bosch is just a showcase of the possibilities ahead, and how the former aims to serve expanded device control capabilities with its software stack.At its CES booth, the company is giving a glimpse of the future. Using the WowMouse app on an Apple Watch, users can play games, summon an AI assistant, use it as an XR input, and control smart home devices.Doublepoint TechnologiesIn addition to the watchOS release, DoublePoint says it has also updated the underlying framework to make some crucial improvements. The static accuracy has shot up to 97%, while mid-walk and running figures have reached 95% and 94%, respectively.This enhanced performance paves the way for innovative applications in smartwatches, fitness wearables, augmented reality headsets, accessibility tools, and a range of everyday use cases, says the company.Apple already offers wrist-based gesture controls on its smartwatch using the Double Tap system. Even though it is quite a refreshing experience, the feature is limited to the Apple Watch Ultra 2, Apple Watch Series 9, and its successor.Editors Recommendations

From Emilia Prez to Shgun, these are the films and TV shows that took home trophies Sunday night.

Sam Altman says his researchers still push back in meetings a positive for him.Altman previously wrote that he is against bureaucracy and supports fostering open communication.Experts emphasize the importance of polite disagreement to maintain a team's trust and efficiency.Sam Altman said that"I spend most of my time with the researchers, and man, I promise you, come with me to the research meeting right after this, and you will see nothing but disrespect. Which is great," Altman said in an interview with Bloomberg Businessweek published on Sunday.His comments echo how the startup's CEO has previously talked about his leadership style.In a 2023 blog post, he wrote: "Fight bullshit and bureaucracy every time you see it and get other people to fight it too." He added,"do not let the org chart get in the way of people working productively together."In a 2019 post on his blog, months before he became OpenAI's CEO, Altman wrote: "One of the best ways to build a network is to develop a reputation for really taking care of the people who work with you." He said leaders should push employees to "accomplish more than they thought they could" without burning out.In the far-ranging Bloomberg interview, Altman also talked about government bureaucracy hindering AI development, returning after he was briefly fired by the board in 2023, and his work schedule.He said his executive team meets for three hours on Mondays. During the week he spoke with Bloomberg, he said he also had six one-on-ones with engineers over two days, a research meeting, and several meetings to discuss "building up compute" and to brainstorm products.He said he communicated far more internally than with people outside the company."I'm not a big inspirational email writer, but lots of one-on-one, small-group meetings and then a lot of stuff over Slack," Altman said.Power of polite disagreementsWorkplace experts say polite disagreement with peers and even top bossesCEOs across tech, including Amazon founder Jeff Bezos, former Google CEO Eric Schmidt, and LinkedIn cofounder Reid Hoffman, have all highlighted the importance of fostering a culture of disagreement from the top.In his 2016 letter to shareholders, Bezos wrote about how employees should embrace his "disagree and commit" strategy, which is a way to say: "Look, I know we disagree on this but will you gamble with me on it? Disagree and commit?""This isn't one way," Bezos added. "If you're the boss, you should do this too. I disagree and commit all the time."Joseph Grenny, a corporate trainer and the coauthor of the workplace strategy book "Crucial Conversations," said staying silent can carry its own pitfalls.In a 2016 interview with the Harvard Business Review, he suggested considering "the risks of not speaking up" which could be the project falling apart or losing the team's trust and weighing those against the consequences of saying something.One way to do it is to ask your manager for permission to disagree, Grenny said, by saying something like, "I know we seem to be moving toward a first-quarter commitment here. I have reasons to think that won't work. I'd like to lay out my reasoning. Would that be OK?"Sabina Nawaz, a CEO coach who was a senior director of human resources at Microsoft for 15 years, wrote in a 2023 LinkedIn post that avoiding disagreement is more damaging to relationships than speaking up. She recommended finding allies for meetings and asking what others think."When co-workers realize you let them proceed with a faulty plan, or waited until the last minute to raise objections, they're likely to lose trust in you," Nawaz wrote.

Sam Altman was ousted as OpenAI's CEO by the company's board in November 2023.Altman said he was left with a "complete mess" after he was reinstated as CEO.Altman said his ouster was "a crazy thing to have to go through" and that he had "no time to recover."OpenAI CEO Sam Altman said the ChatGPT maker was like a house on fire following his brief ouster from the company.Altman told Bloomberg in an interview published Sunday that he was left with a "complete mess on my hands" after being reinstated as CEO.On November 17, 2023, OpenAI's board said in a statement it was removing Altman because he "was not consistently candid in his communications with the board."The board, however, didn't give further details about Altman's firing. Altman was later reinstated as CEO just five days later, after OpenAI's employees protested the board's decision."And it got worse every day. It was like another government investigation, another old board member leaking fake news to the press," Altman told Bloomberg."And all those people that I feel like really fucked me and fucked the company were gone, and now I had to clean up their mess," he added.Altman did not specify which board member he was referring to.Back in November, OpenAI's board consisted of six people: Altman, fellow cofounders Greg Brockman and Ilya Sutskever, Quora CEO Adam D'Angelo, AI researcher Helen Toner, and Tasha McCauley, an entrepreneur and researcher at the RAND Corporation. Sutskever, D'Angelo, Toner, and McCauley had voted for Altman's removal.D'Angelo was the only one of the four to remain on OpenAI's board following Altman's return as CEO. Sutskever left his position as OpenAI's chief scientist in May."It was just a crazy thing to have to go through and then have no time to recover, because the house was on fire," Altman told Bloomberg.When approached for comment, OpenAI told Business Insider that it had nothing further to add to Altman's interview.OpenAI saw multiple exits in its leadership ranks following Altman's return as CEO.Sutskever's co-lead for OpenAI's superalignment team, Jan Leike, left his post at the same time Sutskever did, and joined the company's rival, Anthropic.Then, in August, the company's cofounder and head of its alignment science efforts, John Schulman, left OpenAI to join Anthropic too.In September, OpenAI's CTO Mira Murati announced her departure from the company as well.OpenAI is in talks with California's attorney general's office about becoming a for-profit entity, Bloomberg separately reported. The company was launched as a non-profit research organization in 2015.In October, OpenAI closed a $6.6 billion funding round, valuing it at $157 billion.