The Advocate for Human-Centric AI

Fei-Fei Li is best known for creating ImageNet—the dataset that truly launched deep learning in 2012. As a Stanford professor and co-founder of the Institute for Human-Centered AI (HAI), she serves as a special advisor to the UN Secretary-General.

Vision and Contribution: Her “Human-Centered AI” framework has three concentric circles: individuals (dignity, autonomy, privacy), communities (workers, creators, collaboration), and society (employment, governance, social structures). She advocates that AI should be human-centered, not technology-centered. 🤝

Technical Judgement: While most focus on language models, Li is advancing "spatial intelligence" through her startup World Labs—getting AI to understand 3D space, geometry, and physical context. “We see a lot of beautiful pixel generation, but if you're just generating pixels on a flat plane, they actually lack information,” she explained.

Quote: “Neither 'AI will end humanity' nor 'AI is only utopia' has ever really happened for any technology.” She prefers an analogy with cars: "There were more car accidents 100 years ago than now, but we didn't shut down Ford or GM. We invented seatbelts and speed limits.”