2026-07-29 14:46 UTC
and if i can't elaborate your *theorem statements and specifications* to a logic whose definition i can fit in my head, it's worse than not trusting your proof: i don't even know what it is you want me to believe
Replies (2)
-
@okennedy@discuss.systems 2026-07-29 14:58
@chrisamaphone@hci.social This is the crux of what makes AI so attractive to hucksters. Fitting a regression is easy and gives you a nice pretty graph. Creating a model that is (sometimes) useful is hard and gives you the same pretty graph. Figuring out which type of pretty graph you're looking at is always hard.
-
@carloangiuli@mathstodon.xyz 2026-07-29 16:14
@chrisamaphone@hci.social What about the opposite case, which more or less covers all of ordinary math: logic not too big, but (tower of) definitions extremely big?