🖼️

Fei-Fei Li

Original nameFei-Fei Li

Chinese-American Computer Scientist, Creator of the ImageNet Dataset

Those Who Conceived That "a Machine Could Think" · Pioneers

The ImageNet DatasetThe ImageNet Large Scale Visual Recognition ChallengeAdvocacy of "Human-Centered" AI

Who they are

Fei-Fei Li (1976– ) is a Chinese-American computer scientist long teaching at Stanford University. In 2009 she led the building of the ImageNet dataset, a large-scale visual database of over fourteen million annotated images across more than twenty thousand categories, its construction drawing on tens of thousands of annotators on the Amazon crowdsourcing platform over several years. The ImageNet Large Scale Visual Recognition Challenge (ILSVRC) she then launched gave computer vision a unified, authoritative public testing ground, and in 2012 the AlexNet of Hinton’s team won this competition overwhelmingly, directly triggering the whole AI field’s collective turn to deep learning, a victory inseparable from the data infrastructure Li had earlier built. Li thereafter long advocated the idea of "human-centered" AI research, holding that technological development should always take advancing human welfare and preserving fairness and inclusion as its core goal rather than merely pursuing performance metrics. She is also a co-founder of Stanford’s Institute for Human-Centered AI, and in 2023 published the memoir The Worlds I See, recounting her growth from an immigrant youth to an important founding figure of AI.

Primary sourcesDeng et al., "ImageNet: A Large-Scale Hierarchical Image Database" (2009)Fei-Fei Li’s memoir The Worlds I See (2023)

Key stories

Fourteen Million Images Built a "Testing Ground"

When Li launched the building of the ImageNet dataset, the scholarly world widely held the algorithm model the core of research and the dataset only a supporting role. She held instead that a large enough, diverse enough annotated dataset could itself be infrastructure driving the whole field’s progress — and she proved right: it was precisely this "testing ground" she built that let Hinton’s team’s deep-learning method for the first time prove its power far above the earlier mainstream under a fair, unified standard, triggering the start of the generative-AI wave of the decade after.

Relationships

Disciple

🧠Geoffrey Hinton

Ally

📈Andrew Ng　Both long advocate the idea of "human-centered" AI development, and Ng also once worked with Li at Stanford

Echoes today

Below are how modern works borrow or reinterpret this name or story — not the original material. The two differ, so keep them apart.

The lasting influence of the "data infrastructure" ideaFei-Fei Li’s insistence that "a dataset is itself core research infrastructure" was thereafter proven highly forward-looking, and in the current large-language-model race the importance of high-quality training data is acknowledged to be no less than the algorithm itself.

Appears in

The Deep-Learning Revival and the Shock of ImageNetPersistence Through the "Neural-Network Winter" and the Shock of AlexNet

← Back to Figures