22 - Shard Theory with Quintin Pope
AXRP - the AI X-risk Research Podcast
00:00
The Benefits of Replicating Causal Processes in Deep Learning
Trying to replicate the causal process by which an artifact appeared puts you in a better position to reproduce aspects of the artifact that you didn't know you should be targeting. And I think this is particularly a particularly useful thing to have in the deep learning regime where we have loss functions in SGD. We're not even at the point where anything else can do stuff at all, much less like this stuff we in particular want. So it's like the unknown unknowns that concern me. and I think replicating causal processes which produce those unknown unknowns gives us the best chance of getting close enough along dimensions.
Transcript
Play full episode
