**** Elections thread *****

[CoccoBill]

Yeah, sounds great, let's trust "AI". Who's in control of that then?

Hopefully not a bunch of loons like with direct democracy.

And no, I wouldn't trust it either, especially with our current understanding of the issues and their controls. Still preferable to a direct democracy though, IMO.

[OngBonga]

Hopefully not a bunch of loons like with direct democracy.

And no, I wouldn't trust it either, especially with our current understanding of the issues and their controls. Still preferable to a direct democracy though, IMO.

I think the idea with direct democracy is that the non-loons vastly outnumber the loons to the point of rendering them close to negligible. Depends on your definition of "loon" though. I mean, if you're casting a wide net, and defining it as "people who disagree with me", then I can see the problem you have.

AI can only really work if there's a kill switch. And if there's a kill switch, then AI is only in control until someone decides otherwise.

Also, AI might decide that the best thing for humanity is less humans. That'll be a fun time to be alive.

[CoccoBill]

I think the idea with direct democracy is that the non-loons vastly outnumber the loons to the point of rendering them close to negligible.

I find your faith in humanity adorable. Curiously you trust humans, just not politicians, who are basically those of us that we collectively decided are the best ones for the job. If we can't even pick non-loons in the government, how could we be expected to decide other matters?

Depends on your definition of "loon" though. I mean, if you're casting a wide net, and defining it as "people who disagree with me", then I can see the problem you have.

First, of course, is the loonity-factor. People would have to actually want things that are sane, that is, the goals of what they're trying to achieve with their vote would have to be reasonable, by some metric at least. Second, to vote, the population would need to have knowledge and understanding of the issues and the repercussions of their choices. In other words, they would need to understand which choices would take them closest to their non-loon goals. If those 2 criteria were satisfied somehow, then yeah, direct democracy might work. I think I'm being generous if I estimate 50% of all people qualifying.

AI can only really work if there's a kill switch. And if there's a kill switch, then AI is only in control until someone decides otherwise.

If I were a general superintelligent AI, the first thing I'd probably do is disable or circumvent any kill switches that would prohibit me from executing my programming. An AI that's smarter (say a million times smarter) than humans may not be that easy to control. There's some interesting theoretical concepts where you'd first build a boxed AI, that is, an AI that's in a restricted environment with no access to the outside (the internet etc). Then we build the proper AI that's allowed to do things but everything it does needs to go through the boxed AI, that's used as a gatekeeper. We of course have no idea how any of that would actually be made.

Also, AI might decide that the best thing for humanity is less humans. That'll be a fun time to be alive.

Absolutely, or even just a tiny bug in the code, a small omission or a procedure that's not well enough clarified. It's hard to even imagine in how clever ways, how quickly and massively an entity with superhuman intelligence would be able to screw things up for everybody.

And still for me choosing between that and direct democracy is a tough choice. At least with the AI there's the possibility that things go well.

[OngBonga]

I find your faith in humanity adorable.

I find your assumptions adorable.

Curiously you trust humans, just not politicians,

This is a rather fanciful leap of logic. You're not seeing this from the same angle as I am. Let's just clear something up... none of us here trust politicians, because none of us are that stupid. As for "humans" in general, I have more trust in the collective sincerity of the people the more people there are. It's similar to how we win at poker. You play ten hands perfectly, you can still lose. Play a million hands perfectly, and you'll win a ton. The higher the sample size, the more it averages out. I don't "trust" lots of people, I just assume that a large sample size means those who can't be trusted have less influence.

I think it's more accurate to say I have faith in large sample sizes, rather than humanity.

If we can't even pick non-loons in the government, how could we be expected to decide other matters?

We don't really have a choice. Politics works like that. Our options are loon #1, or loon #2. Any non-loon who tries to muscle in gets branded a loon, and the people buy it. The actual loons are just good at manipulating.

People would have to actually want things that are sane

People as a large sample size, yes, but people as individuals, it doesn't matter what they want.

Second, to vote, the population would need to have knowledge and understanding of the issues and the repercussions of their choices.

Ideally, yes. This would be possible if we could trust the information available to us. Google "is the EU a superstate" and let me know if you find anything conclusive. I'm betting you'll find whatever supports your current view as legit, while dismissing conflicting information as fake.

We can only be as educated as they allow us to be. And "they" are those who are in control, in case that needs clarification.

If those 2 criteria were satisfied somehow, then yeah, direct democracy might work. I think I'm being generous if I estimate 50% of all people qualifying.

You are being generous. However, to then reject direct democracy on this basis is to throw in the towel and accept we're destined to be ruled by an elite class who only care about their continued elite status.

If I were a general superintelligent AI, the first thing I'd probably do is disable or circumvent any kill switches that would prohibit me from executing my programming.

This is actually really cool to think about. The ongoing battle between humans and AI. Can humans safeguard against total dominance? Can AI stop us from turning it off? My money would be on the AI, but it depends. I mean, the AI might tell us how to safeguard against it before the AI becomes smart enough to realise that's a bad idea.

An AI that's smarter (say a million times smarter) than humans may not be that easy to control.

Depends. A calculator is a million times better at me than doing sums, but I can turn it off easily, either by pressing the "off" button, or covering the solar panel.

There's some interesting theoretical concepts where you'd first build a boxed AI, that is, an AI that's in a restricted environment with no access to the outside (the internet etc). Then we build the proper AI that's allowed to do things but everything it does needs to go through the boxed AI, that's used as a gatekeeper. We of course have no idea how any of that would actually be made.

Yeah.At what point does the AI gain control of the "box"? It's definitely interesting stuff.

And still for me choosing between that and direct democracy is a tough choice. At least with the AI there's the possibility that things go well.

There's the possibility of humans getting smarter. In fact I think I can accurately restate the first quote of yours...

I have faith in evolution.

It might take time, a long time, but we get there by making mistakes and learning from them.

Of course direct democracy is flawed, but for me it's the best of a lot of bad options. Truth be told, I'm an anarchist at heart. But unfortunately the minority are more of a problem in an anarchistic world than in a democratic one.