When you say future improvements do you mean bigger GPTs? Improved LLM architectures? Different architectures entirely? Are you sure you're not just imagining a thing that would be scary and then saying "what if it was that?".