Neither chatbots nor AI can do this: the next tech revolution begins - Futura-Sciences

From Chengdu to Silicon Valley: The Journey of Fei-Fei Li Fei-Fei Li, now renowned as the “godmother of AI,” might easily have taken an entirely different path. Born in Beijing in 1976 and raised in Chengdu, China, her life took a dramatic turn at age 16. Moving with her mother to Parsippany, New Jersey, to join her father, she suddenly found herself navigating a brand-new language and culture—a crash course in adaptation that many immigrants know all too well. Her academic path was equally impressive. She studied physics at Princeton University, while also exploring computer science and engineering, earning her master’s degree in 1999. Aiming higher, she completed her doctorate in electrical engineering at the California Institute of Technology (Caltech) in 2005. Her early career included a year as an assistant professor at the University of Illinois Urbana-Champaign, followed by two years in Princeton’s computer science department. In 2009, she became an assistant professor at Stanford, was promoted to associate professor with tenure in 2012, and reached the rank of full professor in 2018—clearly a trajectory not for the faint of heart! Pioneering Computer Vision and Training the Machines Between 2013 and 2018, Fei-Fei Li headed the Stanford Artificial Intelligence Lab (SAIL). There, her research homed in on computer vision, deep learning, and cognitive neuroscience. She was the driving force behind the creation of ImageNet, the now essential annotated image database. Realizing that progress in AI wasn’t limited by algorithms but by the scarcity of data, she set about building the benchmark that has enabled today’s advances in object recognition and computer vision. If you’ve ever marveled at a machine’s ability to “see,” you can probably thank ImageNet. From January 2017, Fei-Fei Li took a nearly two-year sabbatical, joining Google Cloud as Vice President and chief scientist for AI and machine learning. Not bad for a so-called “break.” Questioning Language Models and Today’s Giants Fei-Fei Li is not alone in her critical perspective. Yann LeCun, another luminary in the AI world, is openly skeptical about generative AI models. In his view, they represent a dead end and will never reach artificial general intelligence. Through his company, AMI Labs in Paris, he has just secured nearly €1 billion (over $1 billion) in funding to build AI based on World Models, which many believe could soon revolutionize the field. In 2017, together with her former student Olga Russakovsky, Fei-Fei Li founded AI4ALL, an organization dedicated to promoting diversity and inclusion in AI. Returning to Stanford in the fall of 2018, she established the Human-Centered AI Institute (HAI) in 2019. This new center set out to challenge the tunnel-vision pursuit of performance and profit, raising questions like: “Who benefits from AI, who is harmed, and how can we democratize access?” The Next Wave: Physical AI and Beyond In 2020, Fei-Fei Li became a member of the United States National Academy of Engineering and served on Twitter’s board of directors until the network was acquired by Elon Musk. By 2023, she had joined the United Nations’ Scientific Advisory Board, and in 2025, she received the Queen Elizabeth Prize for Engineering alongside AI luminaries Yann LeCun, Geoffrey Hinton, Yoshua Bengio, Bill Dally, and Jen-Hsun Huang. But the next tech revolution is already underway. After the chatbot craze, artificial intelligence is about to cross a far more tangible boundary in 2026. This budding era is being called “Physical AI,” as industry giants double down on robotics—AI that can truly interact with the physical world. Today, Fei-Fei Li is still a professor at Stanford, but now also spends time in her own startup, World Labs, launched in 2024. She isn’t shy about her views: “Large language models are wordsmiths lost in the dark. They don’t see the real world, and until they understand the world in three dimensions, they’ll remain blind.” With World Labs, she’s exploring large world models (LWMs), AI models capable of understanding three-dimensional space and physical principles. This question is particularly urgent as we enter the age of embodied AI—that is, autonomous robots that need to find their way and interact with the physical world. Forget chatbots’ virtual conversations: the real test of intelligence may come when AI has to get up and move. Or at least, not trip over your coffee table.