This engineer made headlines in May when he resigned from his position at Google to "be able to talk about the dangers of AI" more freely, a few months after the launch of a new version of ChatGPT, a content-generating AI software.
The expert was speaking Wednesday before an audience provided at the Collision technology conference in Toronto, which brings together more than 30,000 entrepreneurs, investors and professionals in the field, most of whom are interested in this innovation and reluctant to listen to warnings about it.
"Before AI becomes smarter than us, I think the people who develop it should be incentivized to work hard to understand how it might try to take control," Hinton said.
"Right now, for 99 very smart people trying to improve AI, there is only one trying to figure out how to prevent it from taking over," he said, pleading once again for serious consideration of the risks raised by this technology.
"People need to understand that this is not science fiction, that it is not just to scare," he insisted. "It's a real risk that we have to keep in mind, and we need to figure out in advance how to manage it."
For the engineer, artificial intelligence also raises fears of widening inequalities.
"The wealth (produced) will not go to the people who work, it will serve to make the rich even richer, and the poor even poorer," he said.
Geoffrey Hinton finally pointed out the danger of false information created by generative software such as ChatGPT.
This is why he calls, while acknowledging that he does not know "if it is technically possible", for an automatic tagging of content designed using this technology, much like central banks do with notes.
© 2023 AFP