
17 - Training for Very High Reliability with Daniel Ziegler
AXRP - the AI X-risk Research Podcast
Is There a Gradient Barrier to Learning?
T seems like either you've got to be adding more data of, like, human labels or something. When things like this, it's actually going to be dangerous. The question is, could your normal training procedure exploit the same information? N: We sort of had the webner facethat i described earlier, where contractors can, can write some things to try to fool a classifier. But we augmented them in a few ways to make their jobs easier.
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.