News for hackers. Stuff that matters.
Re:On the left is GPT-3.5. On the right is GPT-4 by og_kalu on Saturday March 18, 2023 @10:38PM (#35215443)
Except...no lol. Not perfect but Gpt-4's spatial reasoning is consistently solid and if you bothered to test it yourself, it'd be easy to see.

Also Appeal to authority is meh. When these guys can test for this special "true understanding" TM that gpt supposedly lacks then we'll talk. Otherwise it's just vague and ill defined assertions. A distinction you can't test for is not a distinction.

Re:On the left is GPT-3.5. On the right is GPT-4 by jhanschoo on Sunday March 19, 2023 @06:52AM (#35217764)
Depends on the modality of spatial reasoning. I tested it by asking it stroke-by-stroke instructions to write Chinese characters, and it gets something wrong every time and usually in an egregious way.