Raw LLM Responses

What makes you say the gulf between GPTs and true AGI is massive? I mean, we have only a general idea of what kind of emergent structures make the brain what it is. We have no meaningful insight into the inscrutable matrices that constitute the emergent properties of LLMs. Sure, we know they still have architectural limitations, like no persistent state or memory, and no real time feedback loops, that impose significant limitations in adopting the same structures as the human brain, but that doesn't preclude novel structures that can perform similar functions. They don't have to be identical to us in capabilities to qualify as AGI.