Comment by gwern
Thanks for the LLM comment, but that's dumb. If the problem really was as bad with humans (it obviously is not), then OP wouldn't've happened:
> For most of the articles Pangram flagged as written by GenAI, nearly every cited sentence in the article failed verification.
Agree. I'm curious about the human contribution baseline.