How do we know what’s AI-generated vs. sloppy human work? Of course in some situations it is obvious (e.g., video), but text? Audio?

And of course you can even ask AI to add some "human sloppiness" as part of the prompt (spelling mistakes, run-on sentences, or whatever).