“A quadrilateral with perpendicular diagonals is called a kite or a rhombus” yes MAYBE but it could also just be some quadrilateral with no special name. I hate AI so much.
Conversation
Notices
-
Embed this notice
myrmepropagandist (futurebird@sauropods.win)'s status on Sunday, 19-Jan-2025 09:32:22 JST myrmepropagandist
-
Embed this notice
myrmepropagandist (futurebird@sauropods.win)'s status on Sunday, 19-Jan-2025 09:32:18 JST myrmepropagandist
“It sounds kind of correct” is the worst kind of wrong.
GreenSkyOverMe (Monika) repeated this. -
Embed this notice
Matt McIrvin (mattmcirvin@mathstodon.xyz)'s status on Sunday, 19-Jan-2025 09:32:19 JST Matt McIrvin
@futurebird @adriano It's not that it doesn't resemble human thought so much as that it resembles the laziest form of human thought (but after reading an inhuman amount of input).
I know there are mathematicians trying to develop machine learning tools that go beyond the language model approach. But LLMs can't do mathematics worth a damn *because* they are language models.
GreenSkyOverMe (Monika) repeated this. -
Embed this notice
myrmepropagandist (futurebird@sauropods.win)'s status on Sunday, 19-Jan-2025 09:32:20 JST myrmepropagandist
@adriano square is a kind of rhombus I can excuse that. I’m more galled that this is thinking “if A then B” is the same as “if B then A” It’s not helpful. It’s bad. It’s what I spend my days trying to keep them from thinking!
GreenSkyOverMe (Monika) repeated this. -
Embed this notice
Adriano (adriano@lile.cl)'s status on Sunday, 19-Jan-2025 09:32:21 JST Adriano
@futurebird I'm more worried about the definition not including squares.
-
Embed this notice
raganwald 🍓 (raganwald@social.bau-ha.us)'s status on Sunday, 19-Jan-2025 09:33:29 JST raganwald 🍓
@futurebird @mattmcirvin @adriano That's "verisimilitude," and it's what makes Generative AI so awful.
-
Embed this notice