We have no idea about human like/type intelligence. We're not even close to replicating it, we're missing some fundamental things, we don't even know about. https://image.nostr.build/1e457b65c880f6cf...
I don't think I would make any new connections either if my brain state reset after each token. We don't let LLMs be bored and wonder about things while. They spit out tokens in response to a question and then halt.