Neel Nanda on Avoiding an AI Catastrophe with Mechanistic Interpretability
Future of Life Institute Podcast
Mechanistic Interpretability in Deep Learning - Multiple Results
Neil Curry worked on a project called progress measures for groking via mechanistic interpretability. He reverse engineered how his model did modular addition, which turned out to be an algorithm that was very clean and legible. "It's a tiny model doing an algorithmic task. This is the kind of thing we are good at," he says.
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.
