Evidence receipt / uncertainty
Published · transcript-backedPieter Levels: uncertainty
20 Aug 2024 Lex Fridman Podcast #440 – Pieter Levels: Programming, Viral AI Startups, and Digital Nomad Life
“You can ask in the prompt, I don’t know, for example, do you think… Create a ranking score of X Twitter replies where should this post be if we rank on, I don’t know, integrity, reality, fundamental deepness or something, interestness, and it would give you that a pretty good score probably.”
Source trail
Everything needed to verify it.
- Speaker
- Pieter Levels
- Attribution
- Verified speaker
- Claim type
- uncertainty
- Recorded
- 20 Aug 2024
- Publisher
- Lex Fridman Podcast
Transcript context
…So, fact check is a tough one, but it would be interesting to rate a thing based on how well thought out it is and how well argued it is. That seems more doable. That seems like more doable. It seems like a GPT thing because that’s less about the truth and it’s more about the rigor of the thing. Exactly. And you can ask that. You can ask in the prompt, I don’t know, for example, do you think… Create a ranking score of X Twitter replies where should this post be if we rank on, I don’t know, integrity, reality, fundamental deepness or something, interestness, and it would give you that a pretty good score probably. Elon can do this with Grok. He can start using that to check replies because their reply section is chaos. Yeah. And actually the ranking or the reply is not great.…
Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.