But the closest equivalent of a neuron in the AI world would be a node, and I haven't been talking about that. So your comparison does not fit or make sense.

LLMs as a genre of AI research are incapable of reasoning. At best they can pretend to reason, and even then they are terrible at it.