Comment by marliechiller

Personally, I dont think so. I can understand a mathmatical axiom and reason with it. In a sequence of numbers I will be able to tell you N + 1, regardless of where N appears in the sequence. An LLM does not "know" this in the way a human does. It just applies whatever is the most likely thing that the training data suggests.