Evidence receipt / prediction
Published · transcript-backedIlya Sutskever: prediction
27 Mar 2023 Dwarkesh Podcast Ilya Sutskever (OpenAI Chief Scientist) — Why next-token prediction could surpass human intelligence
“We’ve made some progress but much more progress is possible. And so I would expect that ultimately, the thing that will really succeed is when we will have a small neural net that is well understood that’s been given the task to study the behavior of a large neural net that is not understood, to verify.”
Source trail
Everything needed to verify it.
- Speaker
- Ilya Sutskever
- Attribution
- Verified speaker
- Claim type
- prediction
- Recorded
- 27 Mar 2023
- Publisher
- Dwarkesh Podcast
Transcript context
…Do you think that the approaches we’ve taken to understand the model today will be applicable to the actual super-powerful models? Or how applicable will they be? Is it the same kind of thing that will work on them as well or? It's not guaranteed. I would say that right now, our understanding of our models is still quite rudimentary. We’ve made some progress but much more progress is possible. And so I would expect that ultimately, the thing that will really succeed is when we will have a small neural net that is well understood that’s been given the task to study the behavior of a large neural net that is not understood, to verify. By what point is most of the AI research being done by AI?…
Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.