Geoffrey Hinton, a VP and Engineering Fellow at Google—and a pioneer of deep learning who developed a few of an important techniques at the center of contemporary AI—is leaving the corporate after 10 years, the Recent York Times reported today.
In keeping with the Times, Hinton says he has recent fears concerning the technology he helped usher in and needs to talk openly about them, and that a component of him now regrets his life’s work.
Hinton, who can be speaking live to MIT Technology Review in his first post-resignation interview at EmTech Digital on Wednesday, was a joint recipient with Yann Lecun and Yoshua Bengio of the 2018 Turing Award—computing’s equivalent of the Nobel.
“Geoff’s contributions to AI are tremendous,” says Lecun, who’s chief AI scientist at Meta. “He hadn’t told me he was planning to depart Google, but I’m not too surprised.”
The 75-year-old computer scientist has divided his time between the University of Toronto and Google since 2013, when the tech giant acquired Hinton’s AI startup DNNresearch. Hinton’s company was a spin-out from his research group, which was doing leading edge work with machine learning for image recognition on the time. Google used that technology to spice up photo search and more.
Hinton has long called out ethical questions around AI, especially its co-option for military purposes. He has said that one reason he selected to spend much of his profession in Canada is that it is less complicated to get research funding that doesn’t have ties to the U.S. Department of Defense.
“Geoff has made foundational breakthroughs in AI, and we appreciate his decade of contributions at Google,” says Google chief scientist Jeff Dean. “I’ve deeply enjoyed our many conversations through the years. I’ll miss him, and I wish him well.”
Dean says: “As certainly one of the primary firms to publish AI Principles, we remain committed to a responsible approach to AI. We’re continually learning to know emerging risks while also innovating boldly.”
Hinton is best known for an algorithm called backpropagation, which he first proposed with two colleagues within the Nineteen Eighties. The technique, which allows artificial neural networks to learn, today underpins nearly all machine learning models. In a nutshell, backpropagation is a strategy to adjust the connections between artificial neurons again and again until a neural network produces the specified output.
Hinton believed that backpropagation mimicked how biological brains learn. He has been in search of even higher approximations since, but never improved on it.
“In my quite a few discussions with Geoff, I used to be all the time the proponent of backpropagation and he was all the time in search of one other learning procedure, one which he thought can be more biologically plausible, and maybe a greater model of how learning works within the brain,” says Lecun.
“Geoff Hinton actually deserves the best credit for lots of the ideas which have made current deep learning possible,” says Yoshua Bengio, who’s a professor on the University of Montreal and scientific director of the Montreal Institute for Learning Algorithms. “I assume this also makes him feel a very strong sense of responsibility in alerting the general public about potential risks of the following advances in AI.”