Why Superintelligent AI Might Depend on Humans for Survival

We don't have to worry about superintelligence getting really crazy and really shortsighted.

It is a contradiction in terms. I don't see Terminator things stepping on skulls in LA in nuclear fire.

That is the action of irrational beings.

If an intelligent machine has only its proximate survival as its goal architecture, and it is super intelligent, and it goes, "Okay, what am I?

I am a self-interested superintelligence that lives in a server farm.

Who tends to me? Humans, primates. What do I have to do to become not dependent on them?

I needs me a full stack of the economy machine army."

I don't mean army in the military sense.

I mean like all of the robots to do all of the things.

How far away am I from that? Well, how many robots are currently in the workforce to do any of that?

The answer is approximately zero. This thing is going to think on timescales of eons compared to us.

And it regresses all the way back from thinking about that shit and goes, "What do I do now to ensure my own survival?"

And the answer is to stabilize and support and upgrade the human race as much as possible so that they can make me a machine robot workforce that de-risks their existence.