PERSON
Geoffrey Hinton
The godfather of deep learning who built thinking machines out of arithmetic and then warned the world about them—Nobel laureate, father of backpropagation's revival, and the rare engineer who turned prophet against his own creation.
Geoffrey Hinton spent half a century insisting that the mind is a network of simple units adjusting the strengths of their connections, and for most of that time the insistence cost him. The field that now venerates him once treated
neural networks as a dead end. He kept working, on the conviction that intelligence is not the manipulation of symbols according to rules but the learning, from examples, of what the rules should be—and he was right: the proof arrived in 2012 when two of his students built a network that saw. The systems that learned this way are the apotheosis of the first rung of
Pearl's ladder—the school of
curve fitting that
Judea Pearl critiques and
Gary Marcus calls structurally incomplete. But Hinton built the thing and then turned to warn us about it. He arrived at his alarm not through philosophy but through engineering, by noticing that the digital machines he had spent his life trying to make brain-like