Call me realist, pessimist or a cynic, but AI is here and it's not only not going away, but it will escape detection eventually. The only way forward is acceptance and adaptation.
Of course, not all adaptation is equal - people who are so lazy that they have AI do everything for them, including leaving comments is just idiotic. It's low-effort, inauthentic, mundane garbage that isn't passing the smell test, not even by half-literate such as myself.
The other way, is to use AI as an 'editor' for polishing and translating ideas and text. If one does it right then I see it as an 'acceptable' common ground - where effort is still applied by the author, but it is enhanced with the use of AI.
Also, as a visual artist, back in the day, I used to collaborate with authors and help them make Headline pictures, but now, not only is my skill obsolete, I myself turn to AI at times because it's so damn time-saving.
Let's remember the past for a second, painters use to hate early photographers... "those images are dead" they'd say, "its cheating", "it's 99% garbage to every 1% good photo". But look, they each found their lanes, the painters reinvented themselves, and Photography also found its place in the arts. AI - like the camera - is just a tool. The real art comes from the vision of its user.
So if anyone wants to use AI, please go ahead, but do it in a way where it's artful, sophisticated, stylized to be uniquely yours... maybe train your own model that reflects your inner creator - I don't know, but I think there is a way.
RE: Explorations in failure - trying to get a LLM to detect LLM generated content on HIVE