There are huge advantages to AI-moderation. TBD what the unintended consequences are. But I think it's worth trying.
I believe banning AI is a temporary solution. Even today it is very hard to tell human from AI. In the future it will be impossible. We are in the Philip Dick future of "Do Androids Dream" (the book, not the movie). Does it matter if we can't tell human from AI? The book proposes that how we feel about the piece we're reading is the only thing that matter. How the piece got created is irrelevant.
I think what would be nice (but won't happen until cost of AI somewhat decreases):
1. Pre-moderation - AI looks at your comment before you submit it, and suggests changes for clarity, factuality and argumentative strength. You can decide whether to accept these (individual) changes or not. It will also automatically flag if it breaks moderation guidelines too much.
2. Discussion summary - AI will periodically edit main debate points and supporting sources into a comprehensive document, which you can further add to with your comment. This will help to steer the discussion and make it easier to consume in the future. It can also make discussions less ephemeral, which is a huge problem.