What Neel Nanda thinks about metascience
Runs the mechanistic interpretability team at Google DeepMind; worked on interpretability at Anthropic before that.
Neel Nanda did not write this page.
We collected these quotes from things they published elsewhere, and every quote links to where it was said. They have no account here and have not endorsed this site. Quotes are word for word; the short line under each one is our own restatement, not their wording. Their own site. Is this you? Claim it or ask us to remove it. Or tell us what is wrong here.
Filter metascience · Newest first
Only Neel Nanda on metascience. See everyone on metascience
4 dated positions, 2025, in their own words. Our reading of what Neel Nanda has said — not written or endorsed by them.
-
Their wordsthe best kind of interpretability deep dives often involve a bunch of different kinds of evidence all pointing in the same direction. There just isn't really an interpretability technique that I trust enough to see it and then be like, great, I am completely satisfied and have no further questions.
↗How Aligned Is Claude? A Live Review of the Opus 4.5 System Cardyoutube.com 13th of 27 in this recording
-
Their wordsAnd having two different teams both investigate the same thing and see if they reach consistent conclusions seems even better.
↗How Aligned Is Claude? A Live Review of the Opus 4.5 System Cardyoutube.com 14th of 27 in this recording
- 7 days earlier
-
Their wordsUm, so I think that interpretability agents are actually really nice for this kind of thing because the interpretability agents are reproducible in a way that human researchers aren't.
↗Bitter Lesson-Pilled Interp: A Live Paper Review (Activation Oracles & PCD)youtube.com 2nd of 21 in this recording
-
Their wordsa depressing lack of good evidence of things that were done by analyzing mechanisms that could not have been done by just analyzing a model's activations.
↗Bitter Lesson-Pilled Interp: A Live Paper Review (Activation Oracles & PCD)youtube.com 12th of 21 in this recording