Benevolent rule by intelligence:
a thought experiment on the nature of politics

Let us suppose that we are ruled by a superintelligence. From a utilitarian point of view, everything works well.

There is one premise here: this rule has a clear source. It obeys the laws of physics; it cannot create something out of nothing. For example, the weights of a superintelligence run on some server. This step is to separate it from gods or supernatural things that do not obey the laws of physics.

With this condition met, we think there are only two cases of rule. The first is that one or more people are behind it, ruling, or influencing and intervening in, this superintelligence. So in essence it is people who rule. The other case is that there is no one behind it who can directly control it; instead, the superintelligence already has its own judgment, and can run fully automatically.

Accepting the premise above, there is now a lot of discussion about rule by superintelligence, and it is about whether this rule is good or not[1].

But here I propose that the essence of politics is actually the contest and struggle caused by gaps and differences. That is to say, politics cannot be derived directly from utility.

This gap has two sides, quantity and quality, and a change in quantity can cause a change in quality. But take a change in quantity: for example, a society has 1% rulers and 99% ordinary people. The 99% may be like ants, but as the 99% they can still restrain the 1% elite, and politics still exists. Today's democratic countries basically work this way. Once there is superintelligence, the 99% can no longer restrain it. Whether they really can or not, I will not discuss here; I take it as a default premise[2]. But if the difference is one of quality, there is a problem. For example, we raise pigs. Pigs are omnivores and in some cases attack people, but we would not say we are in a political struggle with pigs; we are managing them, raising them. Because we and pigs are different in essence.

If the bad parts of politics are caused by people, because their knowledge is uncoordinated, then what if we replace people with a superintelligence? This is where the thought experiment comes in. The thought experiment is this: you only know that a superintelligence is ruling, but you do not know whether there are people behind it, or whether it is just a superintelligence. Then, for the situation of the people being ruled, is there any difference? In fact, there is none.

It is precisely because of our earlier restriction, that there is a subject that obeys the laws of physics, that the relation between this subject and the ruled constitutes this gap. The chasm of this gap is large enough that, whether there are people behind the superintelligence or not, it all becomes management. At this point politics has already been abolished.

We would not say that God, or some god, is exercising a dictatorship over the world, because that is already something supernatural. What is natural? Natural means obeying the laws of physics. So something that obeys the laws of physics and is also a subject constitutes a kind of control. This control, when the two sides are comparable, or roughly equal, is politics; when they are completely unequal, not in the same category at all, it becomes use or management.

So we conclude that whether people behind it are ruling or a superintelligence is ruling, politics is abolished. Even if values are all solved, the best case is a kind of mercy, and the worst case, of course, is extinction.

  1. Raymond Douglas. "ASI and Liberty: Can we do better than a benevolent god?" LessWrong, October 9, 2026. https://www.lesswrong.com/posts/eSMburFT9eRxHWijf/asi-and-liberty-can-we-do-better-than-a-benevolent-god ↩
  2. Wenjun Cao. "Human Finitude: AI and the Foundation of Civilization." Working paper, SSRN, 2026. https://doi.org/10.2139/ssrn.7476738 ↩