I think it'll be slightly different - without clearly marking AI-generated content, it'll be effectively impossible to find new content that isn't sold to you in pristine packages already, and even that you just sorta have to trust.

Of course, you can't train LLMs on LLM-generated content.