[ericpauley]

That's not OP's contention though; it's that models are incapable of even surfacing sufficient information. This doesn't pass the smell test.

To wit, here is ChatGPT's "superintelligent" answer: https://chatgpt.com/s/t_6a52b4a322bc81919308ec864a0cc056

Seems like a perfectly cromulent description by which a beginner could probably change a tire, perhaps with a few follow-ups to understand what certain components are.

[jamilton]

I read his contention as “the people who wrote AI 2040 are nerds who are so ungrounded from physical reality they wouldn’t even know how to change a bike tire if given instructions”. Which I think is also wrong but has a genuine criticism in it.

[pas]

sure, they have strong bay area aura, but an ad hominem is not very useful criticism of the actual arguments for AI safety (and longtermism, and against x-risks in general).