[stcredzero]

Re: The stagnation of Science and Technology. We've seen such stagnations before in history. Our culture has been losing the idea of science as a process of increasing understanding through inquiry. As expressed by the mass of people, the culture has been going towards a cargo cult worship of science. Western culture once enshrined the erroneous conclusions of Aristotle, instead of engaging in his act of inquiry.

Re: The particular exception of computers. If there were something like a PAIperclip Optimizer, it wouldn't be trying to produce paperclips. Rather, it would be producing something which is highly sought by virtually every person, most of industry, and every sector of society. If there were something like the PAIperclip Optimizer, we would see outsized progress in one particular area, focused around the production of the optimized thing. A super-optimizing AI wouldn't try to kill us off by firing off our missiles. It would instead warp our culture, so that it's producing the optimized thing.

Given that self preservation is such a strong, foundational behavioral pattern, the optimized thing would obviously be more and more faster and faster computers.

[ItsMattyG]

A super-optimizing AI wouldn't try to kill us off by firing off our missiles. It would instead warp our culture, so that it's producing the optimized thing.

This seems like a merely intelligent AI that can't do better than humans at producing the thing. For instance in the Universal Paperclips game (http://www.decisionproblem.com/paperclips/), the AI starts by warping culture around the creation of paperclips.

That is (spoiler alert). . . ...

Until the AI learns to create hypnodrones to get the pesky humans out of the way. Then it uses their matter and the rest of the matter in the universe to create paperclips.

[machinelearning]

The real problem isn't the stagnation of science and technology (in most cases - notable exceptions are maybe healthcare and the climate). The more imminent problem is the economic implication of the embedded growth obligations. And the even higher level problem is that of generations making future obligations they can't meet and don't know about it. A kind of indeterminate optimism.

[machinelearning]

I noticed you downvoted my comment for no specific reason, do you disagree with something I wrote? Or do you misunderstand some part of it? If so I'd be happy to debate/explain it. Thiel himself alludes to this being the mechanism of how stagnation becomes a problem in the podcast, so the source of your discontent is quite unexpected.

[stcredzero]

I did not downvote your comment.

[espeed]

Truth is optimal. A super-optimizing AI would optimize the process of connecting what's true.

[x3n0ph3n3]

The unfortunate thing is that heuristics will achieve 50%, 75%, and 90% of truth with much lower energy costs, and not everyone's energy budget can afford the full cost of truth.