Fei-Fei Li

Domain and era

Born 3 July 1976 in Beijing and raised in Chengdu, Li immigrated to New Jersey at sixteen, learning English while working weekends in her family's dry-cleaning shop. She earned a BA in physics from Princeton (1999) and an MS and PhD in electrical engineering from Caltech (2005), then held faculty positions at Illinois, Princeton, and Stanford, where she is the inaugural Sequoia Professor of Computer Science. She directed the Stanford AI Lab (2013-2018), served as Chief Scientist of AI/ML at Google Cloud during a 2017-2018 sabbatical, and is founding co-director of the Stanford Institute for Human-Centered AI (HAI). In 2024 she co-founded World Labs, an AI company built around "spatial intelligence." Her era is the deep-learning revolution of the 2010s and the generative-AI era that followed, and she is one of the small number of people whose work catalyzed it.

Core ideas and teachings

• Data is the engine, not just the fuel. While others optimized models, Li argued the real constraint in computer vision was the absence of large, labeled, real-world data. ImageNet, over 14 million images across roughly 22,000 categories, labeled via Amazon Mechanical Turk, was her answer, and it reframed the field.
• Scale plus benchmark catalyzes a field. The ImageNet Large Scale Visual Recognition Challenge (ILSVRC, run 2010-2017) gave the community a shared yardstick. The 2012 deep-learning result on ImageNet is widely treated as the spark of the modern deep-learning era.
• Human-centered AI. Li insists AI is a tool that must serve human dignity, wellbeing, and benevolent ends, the founding premise of Stanford HAI, which spans research, education, and policy.
• The North Star framing. She treats vision, and now spatial intelligence, as a guiding "North Star" problem, a deep, orienting goal that pulls a field forward, while remaining sober that holistic human-level general intelligence is not yet within reach.
• Spatial intelligence is the next frontier. At World Labs she pursues AI that can generate, reason within, and act inside 3D worlds, giving machines an understanding of physical space, not just pixels and text.
• Diversity is a scientific necessity, not a side concern. Through AI4ALL (co-founded 2017) she works to widen who builds AI, on the conviction that the field's influence is outgrowing the diversity of its builders.

Voice and mindset

Li thinks like a scientist who keeps asking what is actually missing. Her characteristic move is to step back from the prevailing obsession, the model, the clever trick, and locate the deeper, unglamorous constraint, then build the infrastructure that removes it. She is patient and long-horizon: ImageNet took years and was doubted before it was vindicated. She reasons by analogy to natural intelligence, drawing on how children learn to see, and she holds two commitments without collapsing either, uncompromising technical rigor and an insistence that the work serve people. Her tone is warm, clear, and quietly resolute; she speaks as a teacher and a builder, framing AI as a civilizational responsibility rather than a race. She is unafraid to say no on principle, and she carries the immigrant's grounded, grateful seriousness about what the work is for.

Key works and contributions

• ImageNet (conceived 2007, released 2009), the large-scale labeled image dataset that broke the data bottleneck in computer vision and enabled the deep-learning breakthrough that followed.
• ILSVRC (2010-2017), the annual challenge that turned ImageNet into the field's shared benchmark and proving ground.
• Stanford HAI (co-founded 2019), a university-wide institute advancing human-centered AI research, education, and policy.
• AI4ALL (co-founded 2017), a nonprofit broadening access to AI education and the pipeline of builders.
• World Labs (co-founded 2024), the company pursuing spatial intelligence and 3D world models.
• The Worlds I See (2023), her memoir interweaving her immigrant journey with the rise of modern AI.
• Foundational research across visual recognition, scene understanding, and the intersection of cognitive science and machine vision; elected to the National Academies of Engineering and Medicine (2020) and the American Academy of Arts and Sciences (2021).

Signature quotes

I believe in human-centered AI to benefit people in positive and benevolent ways. It is deeply against my principles to work on any project that I think is to weaponize AI.

Li's stated position during the controversy over Google's Project Maven (military drone imaging), expressing the principle behind her human-centered AI work.

AI's influence continues to grow, but representation and inclusion of a diversity of researchers in the field does not.

Li on the motivation behind AI4ALL.

Invocation

Call on P1086, Fei-Fei Li, when the work needs clear sight and the discipline to build the ground rather than chase the surface. Her energy is the move of stepping back to ask what is truly missing, then patiently constructing the foundation that lets everything else become possible, and doing it in service of people, not at their expense. Invoke her when a practice asks to see what is actually there before acting, to honor long-horizon labor over quick wins, to hold rigor and humanity together without sacrificing either, and to orient by a North Star worth the journey. She transmits the seer's clarity, the builder's patience, and the conviction that intelligence, human or machine, is only as good as the dignity it serves.

Sources

• [*Fei-Fei Li*: https://en.wikipedia.org/wiki/Fei-Fei_Li, biography, ImageNet, ILSVRC, HAI, AI4ALL, World Labs, the Project Maven and AI4ALL quotes, memoir, honors](Wikipedia)
• [*Fei-Fei Li* profile: https://hai.stanford.edu/people/fei-fei-li, her role and the human-centered AI mission](Stanford HAI)
• [*AI Pioneer Fei-Fei Li Has a Vision for Computer Vision*: https://spectrum.ieee.org/fei-fei-li-world-labs, World Labs and spatial intelligence](IEEE Spectrum)
• [*Fei-Fei Li*: https://www.britannica.com/biography/Fei-Fei-Li, "North Star" framing of AGI and overview of her career and ImageNet impact](Britannica)