I mean......this is one of the most well-known ways AI fucks up! It makes shit up! It hallucinates! That's a big reason why it shouldn't be seen as acceptable to just read AI summaries instead of actually reading the sources themselves, because you can't trust whatever the AI shits back out at you. And the bluesky poster is what gets me, like, people go "oh well that's why you have to check AI in your research!" There's still option B: do this kind of research the way everyone did it up until 20-fucking-23 and don't use Claude or ShitGPT.
More takes from the replies:
(this last one gets down to so much AI discourse lmao)




















