Could One Super-Smart AI Ever Rule the Whole World?

An essay published on LessWrong, a popular forum for people who think hard about artificial intelligence, asks a question most of us will never have to answer but all of us may live with: once AI becomes smarter than humans at nearly everything, who runs the planet? The author, Akshay Iyer, frames it as a trap. Either countries stay too divided to prevent disasters, or they hand enormous power to one authority — and absolute power, he warns, always comes with a price.

Why is one ruler so risky? He gives four plain reasons. First, leaders with total power can help themselves and hurt everyone else, and no election can remove them. Second, they get stuck on one idea, even a terrible one — he points to Stalin believing communism would save the world until the day he died. Third, succession goes wrong: the Roman emperor Marcus Aurelius was widely admired, but his son Commodus was not, and Rome suffered. Fourth, mistakes can't be undone, because removing a dictator usually takes force — and an AI-backed dictator would be almost impossible to remove.

What about a single worldwide democracy instead? The author doubts it would hold. His reasoning is simple: AI makes it much easier to run military operations with very few people. In the past, seizing power required convincing generals, soldiers and lawmakers — that friction protected democracy. If AI removes that friction, one determined group could quietly hollow out a democracy from inside. So he wants many countries, each with its own army, acting as checks on each other. If one turns authoritarian, the others can team up and push back.

The honest catch: this is one person's speculation, with no data and no tested plan. He also admits a scary wrinkle — AI systems built in different countries might share the same blind spots or quietly cooperate with each other. And the essay is cut short before it finishes the argument. Treat it as a thought-provoking idea, not a blueprint.