“Progress in AI is inevitable, and is probably good,” field pioneer Geoffrey Hinton said at the Collision tech conference in Toronto on Wednesday. “But we seriously ought to worry about mitigating the bad side effects of it.” (The Logic)

Talking point: Hinton’s interview with The New York Times after he exited Google last month—detailing his fears about the technology he helped develop—brought the existential risk conversation to the mainstream. On stage, the University of Toronto emeritus professor laid out five other categories of potential harms, including one other AI experts have criticized him for overshadowing warnings on: bias and discrimination against marginalized people. That’s a current problem, he acknowledged. “But it is something that I think is relatively easy to fix compared to all the other things,” he said, by targeting not “a completely unbiased system,” but “a system that’s significantly less biased than what it’s replacing.” Systems trained on skewed data can be analyzed and corrected, he said. The other risks included battle robots, fake news and joblessness (which Hinton said can be fixed with “socialism.”)