Isnt this essentially the case by definition? If LLMs can solve a problem it's only because a human already solved that problem (not that this is any different from what humans do)