Even if you're never online and only talk to people in person... over time those people will be increasingly informed by LLM-generate pseudo-knowledge. We aren't just training the AIs. They're training us back.
If you want to live in a society where the people you interact with have brains mostly free of AI-generated pollution, then I'm sorry but that world isn't going to be around much longer. We are entering the London fog era of the Information Age.
We have two and a half generations of people right now most of whom think "I did the research" means "I did half as much reading as the average C student does for a term paper, and all of that reading was in Google."
And Alphabet fiddles while Google burns. This is going to end in chaos.
What's the alternative? No one who says that is saying they did original research, they're saying they searched around and got what they believe to be at least a consensus among the body of experts they trust.
Like I agree the problem sucks but I have no idea what a solution looks like. For fields someone is totally unfamiliar with they simultaneously don't have enough knowledge to evaluate the truth of a claim nor the knowledge to evaluate if someone is qualified and trustworthy enough to believe them. It's turtles all the way down -- especially because topics of any interest you can find as many experts as you care to of whatever qualification you demand making all sorts of contradictory claims.
Is it? Even those whose social life is entirely IRL, they still have to increasingly interact with various businesses, banks, healthcare providers, the government, and often more distant collegues through online services. Do I want these to go through LLM chatbots? No. Can I ensure that I'm speaking to an actual human if the communication is text based? Not really.
The concern isn't necessarily for you. It's for the large swaths of people who are less equipped to filter through noise like this.