I think general-purpose AI might turn out to be one of those technological vanity projects, like eugenics, that we could do, but choose not to.
It is possible, right now, to clone people and genetically engineer them to be "superior". But why would you want to do that? Why would you want a load of identical people going around thinking that they are better than everyone else? Even Fascists shy away from that one these days. But in the twentieth century the Nazis were all over that shit. Times change.
It is possible, right now, to send some guy to Mars. We might even be able to bring them back. But why the fuck would you want to do that? Why would you want to send someone completely insane with half a year of sitting in a tin can getting osteoperosis, break all their bones with a high-G landing on the planet, break them again taking off, then another few months eating shitty food and getting even sicker, also slowly killing them with radiation the entire time? I mean, yeah, we could do it, but why the fuck would you want to spend hundreds of billions of dollars on the world's most expensive home movies? It's nothing more than a convoluted and expensive way of ruining some poor sod's life.
But in the Sixties, the Americans and the Russians were all over that shit. Before Skylab, Salyut, and Mir, it was imagined that zero-gravity was probably a health cure. Before twenty-odd nuclear accidents put pounds and pounds of plutonium and uranium into the air, it was imagined that nuclear power would allow you to spin up artificial gravity anyway. Then we discovered that the Coriolis Effect would make astronauts sick to their stomachs unless you built a spaceship a kilometer or so wide.
And now crewed deep space exploration is understood - by anyone who isn't a hopeless romantic - to be a pissing contest for superpowers, not a serious endeavour. The Moon is just about do-able if you really put your back into it, maybe even some asteroid if you're feeling lucky, but not to live, just to visit, so you can say you've done it. Like climbing Everest.
General-purpose AI and artificial superintelligence might well prove to be the same sort of thing. Yes, we could build a very intelligent computer that knows everything about everything and could run the world like a giant, never-ending game of chess, but why would you want to do that? Why would anyone want to build such a thing?
Narrow AI is easy to quantify. It can win any game, make money, even generate simple computer code, taking the drudgery, scut-work, and general yak-shaving out of a lot of IT jobs. But what kind of fucking idiot would want to put some shitbox in charge of a country, let alone the world? You don't know what's going on inside it, you don't know if it has any subjective experience, but if it does, it's quite likely to fucking hate us. What kind of idiot would build something like that?
