Geoffrey Hinton

person · earth

The computer scientist whose group produced the 2012 ImageNet result that made deep learning the default method in vision, and who is also the author of the most-quoted wrong prediction in the field: that radiologist training should stop because deep learning would supersede human image reading within about five years, said publicly in 2016. Both facts belong in the same record, and keeping them together is most of what this Journey is for. The technical judgement was correct and early; the institutional judgement about what would follow from it was not, and the person best placed to make it got it wrong in the direction of speed and completeness. He has since acknowledged the prediction missed on timing and on breadth. He shared the 2024 Nobel Prize in Physics for foundational work on machine learning with neural networks.

A person is not a place. Drawing it on a map would assert something about the world that no stored fact supports.