Given that LLM content is generated with some randomness, and assuming that published LLM content is filtered to be useful for humans, then even in the case whe

Given that LLM content is generated with some randomness, and assuming that published LLM content is filtered to be useful for humans, then even in the case where no content is being generated by humans, I would expect the published content to contain new information that is generated by humans.