“I don’t think we will see, in the next 10 years, mass worker replacement,” said Cohere co-founder Nick Frosst, because people can still do lots of things that large language models (LLMs) cannot. “I’m surprised you’d say 10 years. I think a few years, maybe,” Nobel laureate Geoffrey Hinton responded. The two AI researchers were speaking at a University of Toronto event on Wednesday. (The Logic)

Talking point: Hinton cited recent reports that graduates are finding it difficult to land entry-level jobs, as companies instead use AI tools to make their existing workers more productive. Frosst—who trained and worked under Hinton before starting Cohere—doesn’t consider LLMs especially human-like, and estimated they will eventually automate about a fifth of desk work.  Hinton, who thinks they’re closer to people, predicted it would be more like four-fifths. Among their points of agreement, both said Canadian businesses have been slow to employ AI, although Frosst said adoption is picking up.