
22 - Shard Theory with Quintin Pope
AXRP - the AI X-risk Research Podcast
The Advantages of Aligning AI to Ambiguity
GPT-4's goal is not to minimize predictive loss like that's the loss function with which it was trained but in so far as it ever has goals they're not to do that unless you like prompt it in some really weird way. So hopefully this is a good way of like us keeping in mind what thing we're supposed to be explaining maybe with shard theory or something else. I'd like to go a little bit back to this idea of like the way we're going to make AI. It seemed like you thought that it would be good to sort of figure out what made humans have human values and kind of replicate that process. Yeah so I kind of like don't think
00:00
Transcript
Play full episode
Remember Everything You Learn from Podcasts
Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.