Oh I hate those chat bots which just display a list of articles matching keywords in your question.

I mean, yeah, but that difference is quite crucial.

People have always wanted to be the top search result without putting effort in, because that brings in ad money.
But without putting effort in, their articles were generally short, had typoes, and there were relatively few such articles.

Now, LLMs allow these same people to pump out hundredfold as much gargage, consisting of lengthy articles in many languages. And because LLMs are specifically trained to produce texts that are human-like, it's difficult for search engines to filter out these bad quality results.

This is exactly the kind of thing that LLMs are good for. I also use them to get quick and concise answers about programming frameworks, instead of trying to triangulate the answer from various anecdotes on stackoverflow, or reading two hours of documentation.

But I figured this kind of thing doesn't count as "slop." OP was talking about the incoherent trash hallucinations, so I left that one out.