Stanford Artificial Intelligence Laboratory (SAIL)

Contribution to AI

McCarthy had already named the field; at Stanford he gave it a building. The lab that opened in 1962 treated artificial intelligence as something to be built rather than argued about, and its choice of problems set the agenda for two decades. Hand-eye work — a camera, an arm, and the question of how one informs the other — turned perception and manipulation into engineering problems with measurable failure. The Stanford Cart crossed a room strewn with chairs in about five hours, which reads as a joke until you notice that every step was a full reconstruction of the scene from stereo images. The lab's other bet was that general reasoning was the wrong target. DENDRAL and MYCIN encoded the judgements of working chemists and physicians as explicit rules, and performed at the level of specialists inside their narrow domains while knowing nothing at all outside them. That result — knowledge beats inference — organised the commercial expert-systems boom of the following decade, and its collapse. SAIL also worked as a dispersal mechanism: its students and its habits of thought seeded robotics groups, vision groups, and the companies that eventually commercialised both.