Evidence receipt / evaluation
Published · transcript-backedPaul Graham: evaluation
9 Aug 2023 Conversations with Tyler Paul Graham on Ambition, Art, and Evaluating Talent
“There was a guy on Twitter — I think his name was Rob Miles — who said that trying to make safe AI will be like trying to make a secure operating system. That is absolutely true, and therefore frightening, because the way you make a secure operating system is not by sitting down and thinking at a table with a piece of paper about the principles for making a secure operating system.”
Source trail
Everything needed to verify it.
- Speaker
- Paul Graham
- Attribution
- Verified speaker
- Claim type
- evaluation
- Recorded
- 9 Aug 2023
- Publisher
- Conversations with Tyler
Transcript context
…Like when it should be modular, or when there should be a regulation on AI is a thing that itself is intractable. I’ll tell you one meta fact though. There was a guy on Twitter — I think his name was Rob Miles — who said that trying to make safe AI will be like trying to make a secure operating system. That is absolutely true, and therefore frightening, because the way you make a secure operating system is not by sitting down and thinking at a table with a piece of paper about the principles for making a secure operating system. More like you try and make such principles, and then someone hacks your operating system. [laughs] Then you think, “Oh, okay, sorry.” You patch it, and then they hack that. Making a secure operating system is like making a fraud-proof tax code. It’s basically a series of patches that were based on successful hacks. Is it Mellon? They say the US tax code is, basically, a series of responses to things Mellon did. Some parts of it, yes.…
Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.