22 - Shard Theory with Quintin Pope
AXRP - the AI X-risk Research Podcast
The Behavior of Deep Learning Models
Deep learning systems seem very broad and ensemble-y which is the word i used previously like people have taken deep models and permuted their different layers. I generally have this notion that deep learning defaults to breadth in a way that would be likeintuitive if you were using common notions of what a utility maximizer would be structured likeCounter: We tend to value lots of different stuff and we sort of asymptotically run out of caring about things when there's lots of themOkay or decreasing marginal value is i guess for any particular like fixed pattern so it kind of broadened the sense of like complex yeah not kind of narrow and simple but they're very complex patterns that are just pointless.
Remember Everything You Learn from Podcasts
