Post #1389487
2026-04-08 07:06 UTC
I lost the source toot, but I read https://arxiv.org/pdf/2604.03437 .
The gist is that in a review of multifarious studies on whether humans can tell ai content from humans', the gist is
- no better than a monkey (to use the snarxiv phrase) on social media posts, cybersecurity, medical reports, sports/fitness coaching, and student essays
- Expert examination like poetry or doctoral candidate screening reliably caught ai.
- Some people are good at it
- AI content is more attractive than humans'
#LLM
Replies (1)
-
@screwlisp@gamerplus.org 2026-04-08 07:21
Historically, a message from a stranger was reliably done by a human whatever their motive. But now messages from strangers (such as essays submitted by students, social media posts, or medical reports) could as well be a GPT. So, community is much more important than it used to be I think. #LLM