News for hackers. Stuff that matters.

A comment on Ask HN: Given LLMs mistakes, do you believe real AI is possible?. Parent comment

Re:Ask HN: Given LLMs mistakes, do you believe real AI is po... by ftxbro on Sunday April 02, 2023 @05:15PM (#35415302)
> anything even close to a common crow's social savvy

GPT-4 has been maxxing out the 'theory of mind' tests. Literally off the charts. They are working on making new tests and charts now.

[ Parent | on HN ]
Re:Ask HN: Given LLMs mistakes, do you believe real AI is po... by skilled on Sunday April 02, 2023 @05:22PM (#35415373)
And do you think it did so well in those tests because it has information about them or because it is intelligent?
[ Parent | on HN ]
Re:Ask HN: Given LLMs mistakes, do you believe real AI is po... by ftxbro on Sunday April 02, 2023 @05:31PM (#35415468)
Well it can't be intelligent because it's only an LLM, so it must be because the test answers are already in its databank.
[ Parent | on HN ]