I didn't say it should have one, I'm saying that LLMs statistically finding the next bit of information aren't really making decisions, which is to counter-argue the fact that it's not different from how we reason.