6 - Debate and Imitative Generalization with Beth Barnes
AXRP - the AI X-risk Research Podcast
00:00
Is There a Problem With Debate?
Sos: How would you gain confidence that, if you have some proticle for debate that seems to work when humans do it? Li: I think mostly the kind of assurances we're going to get o morgan and look something like, well, we can't do any better is more likely than actually saying yes. Sos: So in summery, it seems like your responses like,well, wer can alike, think about the rules really carefully and ere gan a not extromise. But i'm still like confident that it's all fin over or somethingright? Right?
Play episode from 25:42
Transcript
