AI and Existential Risk (Robert Wright & Connor Leahy)
Robert Wright's Nonzero
00:00
The Importance of Understanding Large Language Models
So far as bad actors are concerned, it doesn't matter whether we understand the AI or not, kind of. So solving the so-called interpretability problem, doing a better job of understanding how exactly large language models work is more important for that category. But I think even if like all of AI technology would stop today and no AGI happened, I think this would still be unstable. And we would still see huge new forms of cybercrime and new forms of menstruation. We also don't know how to implement these in computers.
Transcript
Play full episode
