> This argument can be applied to anything LLM generated

That's the whole point, yes. LLMs are inherently unreliable. Both human experts and LLMs are unreliable in their own ways. The human expert has a reputation and some level of responsibility, the LLM doesn't