Evidence receipt / belief
Published · transcript-backedSholto Douglas: belief
28 Mar 2024 Dwarkesh Podcast Sholto Douglas & Trenton Bricken — How LLMs actually think
“I think that's really worth exploring. For example, one of the evals that we did in the paper had it learn a language in context better than a human expert could, over the course of a couple of months.”
Source trail
Everything needed to verify it.
- Speaker
- Sholto Douglas
- Attribution
- Verified speaker
- Claim type
- belief
- Recorded
- 28 Mar 2024
- Publisher
- Dwarkesh Podcast
Transcript context
…In context, are they as sample efficient and smart as humans? I think that's really worth exploring. For example, one of the evals that we did in the paper had it learn a language in context better than a human expert could, over the course of a couple of months. This is only a small demonstration but I'd be really interested to see things like Atari games where you throw in a couple hundred, or a thousand frames, of labeled actions in the same way that you'd show your friend how to play a game and see if it's able to reason through. It might. At the moment, with the infrastructure and stuff, it's still a bit slow at doing that, but I would actually guess that it might just work out of the box in a way that would be pretty mind-blowing. And crucially, I think this language was esoteric enough that it wasn't in the training data.…
Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.