there is at minimum quoted portions of text from the article(s) in question that contain common hallmarks of LLM/AI-generation
I think explicitly describing the heuristic being used by the decision-maker would be better than the presently nebulous "you can tell it's AI because of the way that it is, look at it". The cases with the other surrounding behaviours and especially the change in composition style between post/comments are usually more convincing and the kind of threshold that I'd be after. All I'm figuring is if a discussion thread deserved to exist for the mandatory minimum sentence maybe a thread should exist to determine what that burden of proof should be. Weeklong bans for first infringement on writing that "feels wrong" seems overzealous - there's structure around punishment, there should be as much structure around the assessment of what constitutes rulebreaking
similarly in addition to the excerpts the entire article should be logged verbatim under a collapsible for future reference
How would you recommend motivating authors to write their own material rather than using AI, while balancing the potentially more-demanding ask to reviewers?
the voting module seems to have been pretty effective at kicking AI generated content off the wiki without staff intervention - if there's any examples so far of suspected AI generated content with e.g. more than 20 net upvotes that would be news to me. in practice the heuristic being applied seems mostly of the form "this writing is bad so it must have been AI". reviewers should be able to say "this feels AI generated so I don't want to review it" as an expansion of also being able to say "this writing is so bad I don't want to review it in depth" as separate from any disciplinary processes against the purported author, though it makes perfect sense if it's a pattern of behaviour
Can you provide specific recommendations for these reconsiderations? There isn't really a committee for "ai generation vibe check"
maybe you should explicitly form a committee, or at least a standardised protocol, for the ai generation vibe check. if nothing else, as per the most recent disc threads on AI use, site membership revocation is being used on suspicion of AI generation. if you are confident in the accuracy of the assessment it should be a weeklong ban as per this thread; if you aren't confident then the standard of "suspicion of AI generation may result in revocation of site membership" should have probably come up at some point during this discussion. in recent cases on O5 the explicit weeklong ban on first offense has been de facto replaced with revocation which is counter to the consensus voted on by staff from this thread (and the revocations aren't just a part of a ban as per e.g. here where the user rejoins 2 days later)
so either the mandatory minimum punishment isn't being enforced or the punishment has changed and in either case I feel that warrants further discussion