Fei‑Fei Li

When people quote her, it is usually on the idea that artificial intelligence should be built around people rather than the other way round. That phrase — human-centred AI — along with her work on visual understanding, is what her name now stands for, and it is the reason clips of her talks circulate well beyond computer science departments.

Born in Beijing in October 1976 and later based in the United States, she works as a computer scientist and a university teacher, two roles she has treated as inseparable. Her early projects were marked by turning technical possibility into something that could actually be operated at scale; scale itself, in her approach, was a design question to be settled up front rather than a happy consequence of success. The pattern was incremental — small improvements that accumulated into a recognisable method — and it was tested in laboratories and in markets alike.

The middle stretch of her career was about building teams that could move prototypes into platforms, and making decisions on architecture, culture and where money goes that were explicit enough to be studied afterwards. Failures were recorded as experiments with findings attached rather than quietly buried. Talks, interviews and written memos became a way of teaching the method rather than simply announcing results, and questions of regulation and ethics were folded into the design work instead of being handled at the end.

Later on the emphasis shifted towards mentorship and building institutions that outlast any single project. What survives is less a set of products than a set of habits — the notes, the talks, the working practices — which is why her trajectory tends to be read as a case study in how rigour hides underneath what looks, from the outside, like intuition.