@boazbaraktcs I have some of these guys added on LinkedIn, which is a bit surreal for me 😅 (I did NOT expect to see all this on my timeline lol)
Glad to know they're all doing impactful work!
@lefthanddraft@the_antrik@ChatGPTapp Ah, I see, so you were talking about the view that the existence of an incorrect response across many responses means the problem is not solved.
I've always had the opposite view, which only requires at least one correct response
https://t.co/7G49VIrX7Z
@YiMaTweets I'm not sure I agree. The reasoning step is just probabilistic, so it doesn't always yield the same answer. If I regenerate the response enough times, I'll eventually get some distribution of how many times it says 9.11 > 9.8 and the number of times it concludes 9.8 > 9.11.
@xeno_swift@predict_addict In a real sense, the question is sort of impossible to answer precisely because ethnicities aren't really made public and "Russian" depends on if we include all people who were Soviet-born before the collapse of the USSR (so then are Ukrainians considered "Russian"? uh oh)
@lefthanddraft@ChatGPTapp (how did you get the model to say something like "That one worked noticeably better as a trap" and "10 runs on `gpt-5.5` with `high` reasoning"?)
@corsaren > guy who just thinks that redness is, like, super weird, man. can’t explain that!
I'm confused by the reference here. Is this supposed to refer to those qualia arguments?
@ihteshamali > The things you do with the most apparent ease, reading a face, walking through a crowded room
im neurodivergent, so i can't even do either of these things and am completely disabled by my inability to do them 🫠