> Do we want to encourage the use of LLMs in discussion?

Quite the opposite, but I think we need a better strategy. Downvoting a single offending comment is giving the spammer precise feedback on what worked and what didn't. Bot accounts should just be banned.