That’s the probably most unhelpful and stupid thing I saw in quite some time. The AI or Not Quiz right at the beginning in particular regularly links the exact same things to human or to AI (404 links are a sign or AI hallucination or just natural link decay that show it’s from a human; wonky grammar is a sign for AI unless the text is human-written where it’s just bad english that LLMs would not produce intentionally…).
Yes, spotting AI by clear attributes isn’t easy. If it was those points would be changed first in training of AI. But lengthily discussing tell–tale signs then interpreting them in contrary ways depending on source is absolute bullshit.