I get what he’s saying - like a lot of things he talks about, it only makes sense on a surface level.

A computer is programmed by people, and has assumptions and conditions programmed into it. The people who create the LLM/AI/AGI/Whatever are going to decide what’s important and what isn’t.

That’s literally creating laws and treaties in an overly complicated way, that you probably couldn’t predict the outcome of. It’s never going to be better than having intelligent, educated, and well supported representatives.

Additionally, government doesn’t mean anything unless there’s an enforcement arm. It’s a terrible idea to concentrate the decision making to one thing, then give it the power to enforce its decisions. That’s literally Judgement Day.