I find that a human is able to solve a P=NP situation, and an LLM can’t quite yet do that. When they can the game changes.