Fei-Fei Li

Fei-Fei Li was born in Beijing in 1976, during the final years of the Cultural Revolution. Her early childhood unfolded in a China beginning to open to the world, but her family faced economic hardship. In 1992, at age sixteen, she immigrated with her parents to Parsippany, New Jersey, joining her father who had moved ahead to seek better opportunities. The transition was jarring—she spoke almost no English and found herself navigating American high school while working to help support her family financially.

Li's parents ran a dry-cleaning business, and she spent countless hours after school managing the cash register, doing homework between customers, and translating for her parents. Despite these demands, she excelled academically, discovering a passion for physics and mathematics. She earned admission to Princeton University, where she initially planned to study physics but became increasingly drawn to the intersection of neuroscience, cognitive science, and computation—the emerging questions of how intelligence itself might be understood and replicated.

At Princeton, Li worked with renowned neuroscientist Sebastian Seung, exploring how the brain processes visual information. After graduating in 1999, she spent a year in Tibet as a volunteer teacher before beginning doctoral studies at the California Institute of Technology. Her PhD research, completed in 2005 under the supervision of Pietro Perona, focused on computational models of visual recognition and the fundamental question: how do we teach machines to see and understand the visual world as humans do?

When Li joined the faculty at Princeton in 2007 (before moving to Stanford in 2009), she confronted a stubborn problem in computer vision: existing datasets were too small and narrow to train truly robust visual recognition systems. Researchers had access to thousands of images, but human children learn from millions of visual experiences. Li proposed something audacious: build a dataset mirroring the full breadth of the visual world, with millions of images spanning tens of thousands of object categories, organized according to the WordNet hierarchy of concepts.

The ImageNet project, launched in 2007, was a monumental undertaking. Li and her team, including graduate students and collaborators worldwide, assembled more than fourteen million images, recruiting workers through Amazon's Mechanical Turk platform to label and verify each one. Many colleagues thought the effort misguided—better algorithms, not bigger data, were supposed to be the answer. But Li persisted, believing that scale and diversity were prerequisites for machine perception.

The vindication came in 2012 at the annual ImageNet Large Scale Visual Recognition Challenge (ILSVRC), a competition Li had organized to benchmark progress. A team from the University of Toronto, led by Geoffrey Hinton and including Alex Krizhevsky and Ilya Sutskever, entered a deep convolutional neural network called AlexNet. It shattered previous records, reducing error rates by more than ten percentage points. The result stunned the field. Deep learning, previously seen as promising but limited, suddenly became the dominant paradigm in AI. ImageNet had provided the fuel; neural networks provided the engine. Within a few years, virtually every major technology company had pivoted to deep learning strategies pioneered on ImageNet data.

At Stanford, Li directed the Stanford Artificial Intelligence Laboratory (SAIL) from 2013 to 2018 and the Stanford Vision and Learning Lab, building one of the world's premier research groups in computer vision, machine learning, and robotics. Her research extended beyond static image recognition to video understanding, scene comprehension, and the integration of vision with language and reasoning. She supervised dozens of PhD students who went on to leadership positions in academia and industry, amplifying her influence across the field.

Even as her technical contributions reshaped AI, Li grew increasingly concerned about the technology's societal implications. She observed that AI systems, trained predominantly by a homogeneous group of researchers, risked encoding narrow perspectives and biases. She worried about privacy, fairness, accountability, and the concentration of AI power in a handful of corporations. In 2016, she took a leave from Stanford to serve as Chief Scientist of AI and Machine Learning at Google Cloud, seeking to understand industry challenges firsthand and to advocate for responsible development from within.

Returning to Stanford, Li co-founded (with John Etchemendy) the Stanford Human-Centered Artificial Intelligence Institute (HAI) in 2019, serving as co-director. HAI's mission reflects Li's conviction that AI must be developed with humanity at the center: technology guided by human values, designed to augment rather than replace human capabilities, and accountable to the people it affects. The institute brings together researchers from computer science, law, philosophy, medicine, business, and the social sciences to tackle not only technical challenges but ethical, policy, and societal questions surrounding AI deployment.

Li's immigrant experience and her observations of Silicon Valley's demographic homogeneity galvanized her commitment to broadening participation in AI. In 2015, she co-founded AI4ALL, a nonprofit dedicated to increasing diversity and inclusion in artificial intelligence education, research, and policy. AI4ALL offers summer programs at universities nationwide, targeting high school students from underrepresented and underserved communities—young women, students of color, low-income students—who might never otherwise envision themselves as AI researchers or leaders.

The program provides hands-on AI education, mentorship from researchers and industry professionals, and exposure to the ethical and societal dimensions of the technology. Thousands of students have participated, many going on to study computer science and AI at leading universities. Li's vision is that the generation building tomorrow's AI systems should reflect the diversity of the world those systems will serve, bringing varied perspectives to questions of design, fairness, and impact.

Li has been a prominent public voice on AI ethics, equity, and policy. She has testified before the U.S. Congress, advised government bodies, and contributed to global conversations about AI governance. Her advocacy emphasizes that technical excellence alone is insufficient—AI researchers and developers must grapple with questions of justice, transparency, and the distribution of benefits and harms. She insists that humanistic and technical education must proceed together, that engineers must be versed in history and ethics just as policymakers must understand the technology's capabilities and limits.

Li's contributions have earned her widespread recognition. She has been named to the TIME 100 list of the world's most influential people and has received numerous awards for both her scientific work and her advocacy. In 2020, she was elected to the National Academy of Engineering, one of the highest honors for an engineer in the United States, and to the National Academy of Medicine in 2021, recognizing her work applying AI to healthcare. She has received honorary doctorates and is a fellow of multiple professional societies.

Her research has been cited tens of thousands of times, and ImageNet remains a foundational benchmark in computer vision. Beyond citations, her practical influence is visible across the AI landscape: countless startups, products, and research directions trace their origins to methods developed or validated using ImageNet. Yet Li herself has remained focused less on personal accolades than on shaping the field's trajectory, ensuring that AI serves humanity broadly and equitably.

Li is also an author; her memoir 'The Worlds I See,' published in 2023, interweaves her personal journey—from Beijing to Parsippany to Princeton to Stanford—with the story of AI's recent transformation. The book explores themes of immigration, identity, ambition, and the search for meaning through science. It reveals the human behind the researcher, offering insights into the determination, doubt, and hope that have driven her work.

Fei-Fei Li's legacy is dual. On one hand, she is a builder of foundational infrastructure—ImageNet as the bedrock upon which modern computer vision stands. Her technical vision and persistence unlocked capabilities that had eluded researchers for decades. On the other hand, she is a moral and institutional leader, insisting that AI must be developed with wisdom, care, and accountability. At Stanford HAI, she continues to push for AI research that is not only powerful but also transparent, fair, and aligned with democratic values.

Her work in healthcare AI, applying computer vision to medical imaging and clinical workflows, exemplifies her commitment to beneficial applications. She has collaborated with hospitals to develop systems that analyze patient activity and predict health risks, aiming to improve care quality and efficiency while carefully navigating privacy and consent issues. These projects reflect her belief that AI's greatest promise lies in augmenting human expertise and compassion, not supplanting it.

Li's influence extends to a generation of researchers and students who carry forward her values. The scholars trained in her labs, the young people inspired by AI4ALL, and the policymakers and industry leaders shaped by her advocacy form a network committed to building AI that respects human dignity. In an era when artificial intelligence is reshaping labor, governance, creativity, and even self-understanding, Fei-Fei Li stands as both pioneer and guide—a scientist who helped create the future and who insists that we shape it wisely.