 |
|
|
|||||
Recent breakthroughs in generative AI have centered largely on language and imageryfrom chatbots that compose sonnets and analyze text to voice models that mimic human speech and tools that transform prompts into vivid artwork. But global chip giant Nvidia is now making a bolder claim: the next chapter of AI is about systems that take action in high-stakes, real-world scenarios. At the recent International Conference on Learning Representations (ICLR 2025) in Singapore, Nvidia unveiled more than 70 research papers showcasing advances in AI systems designed to perform complex tasks beyond the digital realm. Driving this shift are agentic and foundational AI models. Nvidias latest research highlights how combining these models can influence the physical worldspanning adaptive robotics, protein design, and real-time reconstruction of dynamic environments for autonomous vehicles. As demand for AI grows across industries, Nvidia is positioning itself as a core infrastructure provider powering this new era of intelligent action. Bryan Catanzaro, vice president of applied deep learning research at Nvidia, described the companys new direction as a full-stack AI initiative. We aim to accelerate every level of the computing stack to amplify the impact and utility of AI across industries, he tells Fast Company. For AI to be truly useful, it must evolve beyond traditional applications and engage meaningfully with real-world use cases. That means building systems capable of reasoning, decision-making, and interacting with the real-world environment to solve practical problems. Among the research presented, four models stood outone of the most promising being Skill Reuse via Skill Adaptation (SRSA). This AI framework enables robots to handle unfamiliar tasks without retraining from scratcha longstanding hurdle in robotics. While most robotic AI systems have focused on basic tasks like picking up objects, more complex jobs such as precision assembly on factory lines remain difficult. Nvidias SRSA model aims to overcome that challenge by leveraging a library of previously learned skills to help robots adapt more quickly. “When faced with a new challenge, the SRSA approach analyzes which existing skill is most similar to the new task, then adapts and extends it as a foundation for learning, Catanzaro says. This brings us a significant step closer to achieving generalization across tasks, something that’s crucial for making robots more flexible and useful in the real world. To make accurate predictions, the system considers object shapes, movements, and expert strategies for similar tasks. According to one research paper, SRSA improved success rates on unseen tasks by 19% and required 2.4 times fewer training samples than existing methods. Over time, we expect this kind of self-reflective, adaptive learning to be transformative for industries like manufacturing, logistics, and disaster responsefields where environments are dynamic and robots need to quickly adapt without extensive retraining,” Catanzaro says. Biotech breakthroughs The biotech sector has traditionally lagged in adopting cutting-edge AI, hindered by data scarcity and the opaque nature of many algorithms. Protein design, essential to drug development, is often hampered by proprietary data silos that slow progress and stifle innovation. To address this, Nvidia introduced Proteínaa large-scale generative model for designing entirely new protein backbones. Built using a powerful class of generative models, it can produce longer, more diverse, and functional proteinsup to 800 amino acids in length. Nvidia claims it outperforms models like Google DeepMinds Genie 2 and Generate Biomedicines Chroma, especially in generating large-chain proteins. According to a paper on Proteína, the team trained the model using 21 million high-quality synthetic protein structures and improved learning thanks to new guidance strategies that ensure realistic outputs during generation. This breakthrough could transform enzyme engineering (and, by extension, vaccine development) by enabling researchers to design novel molecules beyond what occurs in nature. “What makes it especially powerful is its ability to generate proteins with specific shapes and properties, guided by structural labels, Catanzaro says. This gives scientists an unprecedented level of control over the design processallowing them to create entirely new molecules tailored for specific purposes, like new medicines or advanced materials. A new AI tool for autonomous vehicles Another standout from ICLR 2025 is Spatio-Temporal Occupancy Reconstruction Machine (STORM), an AI model capable of reconstructing dynamic 3D environmentslike city streets or forest trailsin under 200 milliseconds. With minimal video input, it produces detailed, real-time spatial maps that can inform rapid machine decision-making. Nvidia sees STORM as a tool for autonomous vehicles, drones, and augmented reality systems navigating complex, moving environments. “One of the biggest backlogs in current models is that they often rely heavily on optimizationan iterative process that takes time to refine and produce accurate 3D reconstructions,” says Catanzaro. “STORM tackles this by achieving high-accuracy results in a single pass, significantly speeding up the process without sacrificing quality. STORMs potential extends beyond vehicles. Catanzaro envisions applications in consumer tech, such as AR glasses capable of mapping a live sports game in real timeallowing viewers to experience the event as if they were on the field. STORMs real-time environmental intelligence moves us closer to a future where machines and devices can perceive, understand, and interact with the physical world as fluidly as humans do, he says. While STORM is built to help machines understand the physical world in real time, Nvidia is also pushing the boundaries of how large language models reasonthrough a project called Nemotron-MIND. This 138-billion-token synthetic pretraining data set is designed to enhance both mathematical and general reasoning. At its core is MIND, a new framework that turns raw math-heavy web documents into rich, multi-turn conversations that mirror how humans work through problems together. By turning dense math documents into conversations between people with different levels of understanding, MIND helps AI models break down problems step by step and explain them naturally. This method doesnt just teach models what the right answer isit helps them learn how to think through problems like a person would. According to its research paper, a seven-billion-parameter model trained on just four billion tokens of MIND-style dialogue outperformed much larger models trained on traditional data sets. It showed significant gains on key reasoning benchmarks like GSM8K (grade school math), MATH, and MMLU (massive multitask language understanding), and achieved a 2.5 percent boost in general reasoning when integrated intoan LLM. Can startups and researchers keep up? Training and deploying advanced AI models requires substantial GPU resources, often out of reach for smaller players. To close this gap, Nvidia is rolling out its next-gen AI models through Nvidia Inference Microservices (NIMs), a suite of containerized, cloud-native tools designed to simplify deployment across different infrastructures. NIM includes prebuilt inference engines for a wide array of models, helping organizations integrate and scale AI with fewer computing resources. Improving efficiency has always been a major focus for us, Catanzaro says. Ultimately, our goal is to democratize access to AI capabilities and make deployment practical at every scale, regardless of their computing resources, to harness the power of AI.” As agentic and foundational AI becomes more capable and more embodied, the future of tech may hinge on how effectively it works with humans. Its critical to identify and support use cases across diverse fields, Catanzaro says.
Category:
E-Commerce
A little over two years ago, AI avatars took the internet by storm as people flocked to apps like Lensa, which generated idealized, often fantastical portraits of themselves. But in the ever-elusive offline world, another, quieter trend has been bubbling up: real portraits, made by real people. Portrait commissions have been on the rise. In 2024, the Royal Society of Portrait Painters, headquartered in London, saw a 40% increase in portrait commissions from American clients now make up roughly 20% of their total. The U.S. has a fascination with the Royal Family more than we do sometimes, says Martina Merelli, fine art commissions manager at the Royal Society of Portrait Painters. Its an acknowledgment of the quality of work. It’s no wonder Americans are fascinated. Since its founding in 1891, the Royal Society of Portrait Painters, also known as RP, has been the society of choice for the British Royal Familys public and private commissions. Its members have famously painted the late Queen Elizabeth II, Princess Diana, the Duke of Edinburgh, the Prince and Princess of Wales, and Prince Harry. Notable figures like Dame Judy Dench, Sir David Attenborough, and Stephen Hawking have also been captured on canvas. [Image: Courtesy of Ben Sullivan/the Royal Society of Portrait Painters] But the commission service isn’t limited to the elite. As long as you have disposable income (a head and shoulder begins at around $6,500) anyone can commission a portrait. At a time where AI is squashing many artists, this particular art form is enduringperhaps as a symbol of our need for tangible human connection. A brief history of portraiture Portraits, like art more broadly, have long been seen as a mirror to society. Before the camera was invented, the only way to record someone’s likeness was to paint, or sculpt, a portrait of them. But portraits were never just a recordthey were signifiers of wealth, taste, and power. In ancient Egypt, painted portraits were placed over mummies to guide them into the afterlife. In Ancient Rome, they were used to commemorate the dead and assert lineage. Emperors used them to reinforce authority. Dictators turned them into propaganda tools. One of the first portraits to depict a merchant couple from the middle ranks of society appeared during the Renaissance, when the focus expanded from rulers, nobility, and clergy, to wealthy merchants, bankers, and scholars. Today, portraiture remains intertwined with global politics and economic tides. Its no secret that many of our clients are brokers, bankers, hedge fund managerspeople whose decisions are deeply affected by how the market is going, Merelli says. In 2024, the U.K. saw two major elections. These ushered in a transition from a conservative to a Labour government that directly impacted the tax structures around private schooling. Merelli recalls one acquaintance with three daughters in private schools remarking that taxation money used to be their art money. The new faces of portraiture [Image: Courtesy of Frances Bell/the Royal Society of Portrait Painters] Despite its exclusive history, over the past few decades, the art of portraiture has become more accessible. Frances Bell, an RP member who has been painting portraits for over 20 years, says her clientele now includes newlyweds, young professionals, and parents wanting to leave behind a tangible heirloom. Its a time stamp, she says. Something important they will carry on. Institutional portraits of CEOs, lawyers, chancellors, and the like still account for a big portion of the market. (Bell has also painted members of the royal family but these are cloaked in NDAs.) She believes the impulse behind a portrait commission often goes deeper than vanity. “I’m not saying it’s not there, I think it’s there for all of us, but I get people who want a little thrum of the life force to be put on into the canvas to last forever,” she says. “It’s that feeling of posterity, and permanence.” Unsurprisingly, that kind of posterity doesnt come cheap. Merelliwho often acts as “cupid” between prospective clients and painters at the RPsays the average price for a portrait in 2025 has decreased from what it used to be, but it still hovers around $13,000$20,000. “You can go up to $130,000 depending who the artist is, what brief you have, but a comfortable number is probably $66,000 to $80,000 if you want a full length of yourself with your house in the background and the dogs.” (Frances, who was trained at the prestigious Charles. H. Cecil Studios in Florence, charges $10,000 and upwards for a head and shoulders paintingor about $4,000 for a charcoal drawing.) A proud antithesis to AI That portraiture remains popular is both a rejection of the zeitgeist and, paradoxically, a natural extension of it. It is a slow process that can take countless hours over many sittings, and that i precisely why it is appealing. “It’s quite confessional,” says Bell, who places great importance on the in-person sittings. “I have their secrets coming out of my ears.” Everyone interviewed for this story emphasized the intimacy of the sitting process. Something about two people breathing the same air, in the same room, and looking at each other for hours. For Anthony Connolly, president of the Royal Society of Portrait Painters, this dynamic even shapes the vocabulary painters use. While photographs shoot, painters find a presence, come to a lightness. “You’re there, with your model and it’s like a triangular conversation, where the third point of the triangle is the thing you’re making,” says Connolly. The connection goes both ways. For the painter, its an act of seeing. For the sitter, its an act of being seen. Its a bonding experiencean art formthat no algorithm can ever replicate. An investment piece Claudia Fisher, an American who moved to the UK around the beginning of the pandemic, was not allowed to divulge the cost of her paintinga head and shoulder by painter Paul Brason. Having never owned a piece of art before, the cost was “one giant gulp,” she says. But she has no regrets. [Image: Courtesy of Paul Brason/the Royal Society of Portrait Painters] Fisher, now 69, was reading a book about the social history of tiaras when I called her. After a multifaceted career as an opera singer and a classical architecture designer, she has turned to fashion and today runs a fashion label called Belle Brummell, which makes luxury jackets inspired by 18th and 19th century British couture. Fisher wanted her portrait to act as a marketing tool for her designs. She had just wrapped up the first prototype of her jacket, when it dawned on her: what better way to evoke the historical spirit of her brand than to be portrayed in one of her own designs, in a composition reminiscent of the era? “I’ve always loved the idea of getting a portrait done because I had vision of myself being in a gorgeous dress,” she says with a laugh. “It wasn’t about immortalizing me, I just wanted a pretty dress.” She got a pretty jacket instead. Fisher made four separate trips to Bath, where Brason lives, on four separate occasions. Brason also traveled to her and her husband’s house in Brighton to get a sense of her personality at home, take reference photos, and do a pencil sketch. The two are still in touch. “If were in the area Ill call and see if hes around,” she says. “These relationships continue.”
Category:
E-Commerce
When the electric car startup Rivian was set to release its first vehicle in late 2021, the company made the unconventional choice. Instead of a more conventional neutral tone, it manufactured a significant amount of its initial production run in a custom color the company called Launch Green. It was a decision that ran counter to almost every color trend and automotive industry sales report, and one that’s come to shape the way the company builds out one of the most unique color palettes in the car business. “Everybody buys black, white, or gray. Pretty much every single brand, they’re going to have that. And it doesn’t matter if you’re in the U.S., you’re in China, you’re in Europe, that’s what it is,” says Jeff Hammoud, Rivian’s chief design officer. “Those are the ones that people order. But they’re not the ones that create the most buzz or excitement.” Launch Green, marketed as a limited run for the company’s R1T truck, bucked the trends and rose near the top of color rankings among Rivian fans and buyers. The Rivian forum on Reddit had such a heated debate over Launch Green’s merit that its moderator pinned a note to the top of the comments thread stating that it had been reported by some users for “incorrectly” placing Launch Green in second place. “I understand that many of you feel personally victimized by Launch Green not being #1. I encourage you to take a break from the internet or talk to a loved one for support,” the note read. Though the company doesn’t break down its sales figures publicly, Launch Green was immediately popular. Despite being a limited run, customers still ask for it nearly five years later. [Photo: Rivian] Colors that look good dirty Considering the approach validated, the company has since put an uncommon amount of effort into its color palette, not only creating unique custom colors but also making those colors an extension of Rivian’s adventure-centric, California-inspired brand. From L.A. Silver to El Cap Granite to Red Canyon to Storm Blue, Rivian’s paint options purposely lean into an outdoorsy theme. The company just announced another limited edition paint and trim package, California Dune, a pale sand color that evokes off-roading in the desert. [Image: Rivian] “We wanted something that like looked crisp and clean and premium,” Hammoud says. “If it’s dirty you can’t really tell. It’s not this car that you feel like you constantly have to clean, like a black car.” Rivian does offer its vehicles in black”Midnight” in the company’s parlancebut only reluctantly. Hammoud says Rivian CEO RJ Scaringe doesn’t like black, which he finds to be “not an optimistic color” and one that’s not exactly on brand. “But,” Hammoud says, “customers love it.” For some, Rivian’s colors may just look like slightly different versions of a blue or red that any other car company might use. But according to Hammoud, Rivian’s colors have been carefully developed to reflect a spirit of adventure, while also being bespoke to the brand. “We want it to have that warmth that our brand has, and also something that invites you to get it dirty,” he says. [Photo: Rivian] That approach to color has become so ingrained in the brand’s approach that Rivian hired its own in-house paint specialist, enabling it to develop new color options faster. Even so, adding new colors to the palettethere have been 12 so farrequires a significant investment of time and coordination with suppliers. “The fascias, the mirrors, the door handles, parts of the liftgates, none of those are actually painted at our plant,” Hammoud says. “So we have to work with all these different suppliers to essentially take that same color and make sure it matches identical.” [Photo: Rivian] Adding a new color can take years, but Hammoud says that limited color runs can happen much faster, since the company’s manufacturing facility in Normal, Illinois, can swap a color into the production line for a short time before returning to a more standard color. Bringing a new color like California Dune into the lineup for a limited run is another way for the company to generate some brand buzz. “It’s a fun and I wouldn’t say easy but a light lift for us to be able to go and add freshness to the vehicle by offering a new color,” Hammoud says. [Photo: Rivian] Rivian is also careful about when to take a color out of the lineup. One discontinued color, Compass Yellow, had consistently high Net Promoter Scores, a measure of how likely a customer is to recommend a product to others. “People were the most passionate about that color and Red Canyon, which are really low take rates for us,” Hammoud says. Though the yellow was dropped from the lineup, the red is still available. [Photo: Rivian] These color choices are partly driven by sales figures and customer demand, but Hammoud says the company’s overall approach to color is more closely tied to the adventurous image it’s trying to create with its off-road-ready truck and SUV models. The company pays attention to color trends in the automotive world, but isn’t concerned with simply keeping pace with competitors. “Everything we do from a color standpoint is influenced by the types of products that we think align with our brand, align with our customers. And a lot of that starts from outdoor adventure gear, footwear, backpacks,” Hammoud says. This extends to other sides of Rivian design, like the brand’s distinctive headlights, which were inspired by a rock climbing carabiner. [Photo: Rivian] But Hammoud says color may be one of the most important elements of Rivian’s vehicles. “Color is a big part of purchase consideration for people,” he says. Ultimately the cars are products, and the company is trying to sell them. Color, he argues, helps make the cars more distinctive, which leads to more customer interest, and maybe a foothold in a crowded marketplace. “Finding inspiration from outside of automotive is a big part of it,” Hammoud says. “If you don’t do that, you’re just going to feel like you’re every other car brand.”
Category:
E-Commerce
All news |
||||||||||||||||||
|
||||||||||||||||||