This statement does not logically cohere. "We can spot it because so many people make it easy to spot." You don't see how this is just petitio principii in action?
To fully disguise LLM prose, you'd have to rewrite it entirely, and if you were going to do that, you wouldn't be the sort of person to use it in the first place.
Maybe some people stop reading LLM slop purely because it violates their moral principles or whatever but most people bail out because slop is mentally painful to read. If you are a human and you write like today’s AI find a different writing style, not because reads like AI, but because it reads like shit.
LLM writing, unless someone puts in the time to improve it, typically follows the exact same patterns and favors the same words. You can’t reas a thread here without people talking about “Claude speak.” It is readily apparent, I do not need to show you a 10 year study with n=100,000 to make this point. The article isn’t hallucinating a problem, we are all nodding along because we all see it every goddamn day lmao. He even cited a study and makes a pretty strong case for why it’s a useful metric here in TFA.
If the AI writing is indistinguishable from human writing, then it is not lazy copy and pasting of AI outputs and isn’t just raw LLM output with no work done on it. So in that case it’s no longer a problem.
LLM’s cannot write in a natural, human way that distinguishes it from the typical LLM output on the first try. If they could, we wouldn’t have this problem. Maybe one day they will. Hell maybe it’ll even be next week. But currently they do not so I do not understand why we are having this discussion.