[NBJack]

Strangely enough, it is actually solid proof that they have a mental model; they just reject your attempts to amend it. This is where many assert several fundamental decisions of that model are actually more driven by the heart rather than the brain.

A LLM has no such concept. It is not truly capable of accepting your rational arguments; it may nod emphatically in its response after a few sentences, but once your input leaves its context window, all bets are off. We can retrain it with your particular conversation, and we will increase likelihood it will more readily agree with you, but that is likely a drop in the bucket of the larger corpus.

But this is well past technology and into philosophy, of which I can safely say my mental model for the topic would likely be rather frustrating to others.