Hinton will speak at EmTech Digital on Wednesday
Linda Nilind / eyevine via Redux
Geoffrey Hinton, Google vice president and engineering fellow and deep learning pioneer who developed some of the most important technologies at the heart of modern AI, is leaving the company after 10 years, New The York Times reported today.
According to The New York Times, Hinton has new concerns about the technology he helped deploy and wants to talk openly about it, and part of him is now doing his life’s work. says he regrets it.
Speaking live to MIT Technology Review in his first post-retirement interview Wednesday at EmTech Digital, Hinton shared the 2018 Turing Award (computing’s equivalent of the Nobel Prize) with Yann Lecun and Yoshua Bengio.
Don’t settle for half the story.
Get paywall-free access to tech news today.
subscribe now
Already a subscriber?login
MIT Technology Review provides an intelligent, independent filter for the flood of information about technology.
subscribe now
Already a subscriber?login
Lecun, Meta’s lead AI scientist, said: “He didn’t say he was going to quit his job at Google, but I wasn’t too surprised.”
The 75-year-old computer scientist has split 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 at the time was doing cutting-edge research in machine learning for image recognition. Google has used its technology to power things like photo search.
Hinton has long raised ethical questions about the shared use of AI, especially for military purposes. He said one of the reasons he chose to spend most of his career in Canada was because it was easier to get research funding unrelated to the U.S. Department of Defense.
Hinton is best known for an algorithm called backpropagation, which he first proposed with two colleagues in the 1980s. This method of training artificial neural networks underpins nearly all machine learning models today. Simply put, backpropagation is a method of repeatedly adjusting the connections between artificial neurons until the neural network produces the desired output.
Hinton believed that backpropagation mimics the way biological brains learn. Since then he has been looking for better approximations, but no improvement.
“In my many discussions with Jeff, I was always a proponent of backpropagation. He thought it was a better model of how it would work,” he says. Lekun.
Yoshua Bengio, professor at the University of Montreal and scientific director of the Montreal Institute for Learning Algorithms, said: “I think this also makes us feel particularly responsible in warning the public about the potential risks of subsequent advances in AI.”
MIT Technology Review will be sharing more about Hinton throughout the week. Don’t miss Will Douglas Heaven’s live interview with Hinton on his EmTech Digital on Wednesday, May 3rd at 13.30 ET. with ticket From the event site.
