I think there has already been people on this board admittedly using LLMs to generate comments. This theory predated this current AI explosion, but that potentially makes it harder to tell if you are arguing with a person or a bot.
In contrast, if you join a curated group, and you yourself curate the groups you join, a lot of the guards people put up when they hop onto e.g. Twitter can be lowered and people are more interested in engaging in good faith. There is still some work to be done as regards flow of information between federated instances to promote substantive discussion while avoiding filter bubbles, but it's a better start than the simplest kind of flattening of society that is expressed today in Twitter and the like.
----
The above was generated by ChatGPT.