Evidence receipt / commitment
Published · transcript-backedGwern: commitment
13 Nov 2024 Dwarkesh Podcast Gwern — Anonymous writer who predicted AI trajectory on $12K/year salary
“Then, as I read more and more of the meta reinforcement learning literature, from DeepMind especially, I added in material about neural networks.”
— Gwern
Source trail
Everything needed to verify it.
- Speaker
- Gwern
- Attribution
- Verified speaker
- Claim type
- commitment
- Recorded
- 13 Nov 2024
- Publisher
- Dwarkesh Podcast
Transcript context
…Once you notice one example of this pattern, like this pain example, do you just keep adding examples to that? Walk me through the process over time. For that specific essay, the first versions were about corporations not evolving. Then, as I read more and more of the meta reinforcement learning literature, from DeepMind especially, I added in material about neural networks. I kept reading and thinking about the philosophy of mind papers that I had read. I eventually nailed down the idea that pain might be another instance of this: “Pain makes us learn. We can’t get rid of it, because we need it to keep us honest.” At that point you have more or less the structure of the current essay. Are there examples where it’s not a matter of accumulating different instances of what you later realize is one bigger pattern? Rather, you just have to have the full thesis at once.…
Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.