You’re questioning these things but i’m not offering an opinion, I’m explaining

You’re questioning these things but i’m not offering an opinion, I’m explaining the defacto mechanism by which the brain performs its functions. Now I can explaining with higher resolution at the cost of much more work on my part, because I understand it to the molecular level – at least where it matters.

There is a sort of interesting utility in the means by which GPT is creating tokens and vectors and using grammar that is rather elegant. But this remains a process of error reduction and probability of word choice improvement … not one of ‘understanding’.

LLMs don’t and can’t understand without a world model.