[NooneAtAll3]

> and I've since checked with an LLM

but did you recheck it yourself, or are you trusting unreliable narrator?

[matthewdgreen]

I only checked the abstracts, and they seem consistent. Good LLMs (Claude Opus, ChatGPT Pro) still get things wrong regularly, but lately I've noticed these are mainly the deep details, not easy things like "there is a result that claims X."