Explain yourselves, AI doomsayers

AI might end the world as we know it, or so the refrain so often goes. But what exactly does that mean?

When Geoffrey Hinton, the "godfather" of neural networks, quit what must have been a pretty sweet gig at Google, he did so to openly warn about the pace of Al development. He's not alone: 1,000 industry players signed a letter demanding a six-month developmental moratorium; Sam Altman, CEO of GPT-maker OpenAI, admitted that his own company is "a little bit scared of this"; controversial academic Eliezer Yudkowsky suggested we should preemptively bomb data centres to counteract the AI threat.

Unlike the others, Hinton has considered in some detail how AI could hurt us. In the short term, he worries about AI flooding the internet with disinformation, making trust impossible. Given that's already happening, it's a fair concern.

He's also worried about what happens next. I'm not going to put words in the mouth of an academic legend, so here's what he told the New York Times: "Look at how it [AI] was five years ago and how it is now," he said. "Take the difference and propagate it forwards. That's scary." Assuming he's right, AI will soon outsmart people. Hinton sets this milestone 50 years into the future, giving us time to prepare, presumably.