Fear and policy in San Francisco: Anthropic and the AI reckoning
In an exclusive interview, the philosopher tasked with teaching Claude right from wrong talks safety, identity, and whether the ceiling on AI capability is closer than we think.
AI arguably grinds on the fumes of hype, so it can be hard to know when to pay attention, and when to scroll past the dystopian doomsayers.
But for Amanda Askell, Anthropic’s resident philosopher and in house ethicist, what looks like hype or fear-mongering has in the AI era more often than not turned out to be prescient.
“I remember when GPT-2 came out, I said people might need to start taking exams in pen and paper again because students will use AI to help with their homework, and lots of people were like, ‘you’re just hyping up these systems,’” Askell tells Capital Brief in her first Australian interview.
“In retrospect, this isn’t now considered a wild view. Yesterday’s hype is in many cases just hype from a year ago — today it’s just, oh, that’s the norm.”
