Once LLM-powered bots flood social media [1], websites will have to implement verification as a means of survival.
[1] That hasn't even really begun. Wait until ChatGPT+ capabilities can be run from home at next to zero cost.
If people generate better content with bots, that's a good thing. If the content people generate with bots is worse, then they're not going to be successful with it.
If bad content outperforms good content, that would be a problem for the social network. ... but that would be a problem regardless of the tools used to create that content.
How would my position in this argument be any less valuable if it turned out that I automated it with the help of a text generator?
I wouldn't personally act on this, of course. It's not at all constructive. I find it amusing to contemplate, though, and I also believe it's plausible others will execute on a similar vision.
There will be all sorts of other actors using LLMs for their own ends, too. Astroturfers. Trolls. Griefers. You name it.