Not sure if this is the right place to post about this, but it’s the closest topic I could find.
Recently we are seeing an uptick in AI generated posts; new users with complicated wall-of-text posts, a bit too well written. The smoking gun is similar posts appearing on a different Discourse forum.
My suspicion is a user previously banned for using AI in obnoxious ways, returning to test their latest model. These new posts are of much better quality than previous efforts. The next generation of AI generated posts is going to so hard to distinguish from human writing, we may never be know the difference. The question is, what can be done about it, if anything?
So here is a suggestion, possibly a BAD suggestion, but here it goes anyway … We need our own AI-Responders to respond to suspicious new users, answering obvious questions and testing for human generated responses. It might need to happen in a sort of sandbox so it would not be obvious to the new user they are being tested.
Original credit for the idea to Randall Munroe.