this post was submitted on 15 Dec 2025
628 points (97.0% liked)

Science Memes

17750 readers
1885 users here now

Welcome to c/science_memes @ Mander.xyz!

A place for majestic STEMLORD peacocking, as well as memes about the realities of working in a lab.



Rules

  1. Don't throw mud. Behave like an intellectual and remember the human.
  2. Keep it rooted (on topic).
  3. No spam.
  4. Infographics welcome, get schooled.

This is a science community. We use the Dawkins definition of meme.



Research Committee

Other Mander Communities

Science and Research

Biology and Life Sciences

Physical Sciences

Humanities and Social Sciences

Practical and Applied Sciences

Memes

Miscellaneous

founded 2 years ago
MODERATORS
 
you are viewing a single comment's thread
view the rest of the comments
[–] ryathal@sh.itjust.works 111 points 3 days ago (11 children)

You missed what they meant. It means gpt5 is really good at one arbitrary and extremely specific topic. Anything else it's comparable with a random person on the street.

[–] pennomi@lemmy.world 45 points 3 days ago (8 children)

Reality is the opposite though. GPT5 is expert in a pretty wide amount of trivia. It’s better than the average uneducated person in every subject, but worse than an expert in every subject.

[–] finitebanjo@piefed.world 5 points 3 days ago* (last edited 3 days ago) (3 children)

All the AI are wrong at least 1 in 20 tokens, probably more like 3 in 20, to their sample data which is also wrong a considerable amount of the time, so it is an expert in literally nothing and never will be.

That 1 in 20 number comes from the 2020 paper by OpenAI and the 2022 paper correcting it by Deepmind about AI scaling laws where the AI inverse error rate would never reach 94% with infnite training and power.

load more comments (2 replies)
load more comments (6 replies)
load more comments (8 replies)