I think it's inevitable that every internet community will be overtaken by AI chatbots. While that may initially be a problem for social media platforms that rely on people watching ads, they will capitalize on an even bigger market- advertising directly to the bots themselves. When bots gain functionality to make purchases on behalf of their user, advertisers will compete to influence them.
People will grow so trusting of their AI agents that they will allow them free reign to make decisions on which products or services to buy, so they will in effect become the primary decision maker of that household. And that's who advertisers target. By then, the advertisers themselves will be mostly AI guided as well, so it will be AI influencing AI, all at hyperspeed through countless iterations.
I shudder to think how this world will look, but it will happen sooner than we expect.
I had this thought a couple weeks back when I was a little bit high:
What if humans develop non-sentient AI like ChatGPT in the form of humanoid robots, but then humanity dies off before real sentient AI happens? So, you have a whole planet full of what are basically chat bots pretending to be people and doing everything they used to do, and copying from one another as opposed to receiving any new data/programming from humans.
Ninety percent of them will be dead from lack of hardware maintenance before anything interesting happens.
Extremely weird that there was no overlap between the "humanity can build durable computer hardware" era and "humanity can build a passable non-sentient AI" era.
Yeah it's not realistic at all, but could make for an interesting story setting if someone could resolve those major holes.