Saying "oh some future AI will solve it" doesn't answer those fundamental questions about why we should trust an AI government over anything else. You're declaring some perfect AI by fiat without addressing the actual problems with the proposal.