Geoffrey Hinton

If you have come across a clip of Geoffrey Hinton talking, it is probably about neural networks — the approach to machine learning that his name has become shorthand for, alongside representation learning. He is a British computer scientist and artificial intelligence researcher, born in London in December 1947, and much of what circulates from him comes not from papers but from the talks, interviews and public remarks he has used to explain how the work is actually done.

The through-line in the accounts of his career is method rather than flourish. Early on he treated scale as something to be designed for rather than hoped for, and the record describes a long run of small improvements that eventually hardened into a recognisable way of working. Failed attempts were logged as experiments that taught something, not tidied out of the story. Later, running teams that turned prototypes into platforms, the same habits showed up in decisions about architecture, culture and where money went — and in the view that ethical and regulatory questions belonged inside the design problem, not outside it.

In his later years the emphasis shifted towards mentorship and building institutions that could outlast any single project. That is part of why he draws the audience he does: the interest is less in a particular result than in the standard of execution behind it, and in whether that standard can be reproduced by other people in other fields. His academic page sits at cs.toronto.edu.