[Boudica]

Stephen hawking rambles on about the dangers of AI (his field is physics so yeah, while he is intelligent not really his domain so to speak)

Then you have got Ex Machina, That one with Johnny Depp (can't remember what it's called but it's recent) you've also got Hal and of course most recently you have got Halo 5 which has always featured bat shit insane AI.

Pretty sure it's Transcendence owtte? Those are just some of the recent stuff. I remember watching AI: Artificial Intelligence (that film from ages ago) and it made me cry over a robot. That covered so many questions about how AI would fit in with humanity. I feel like it's probably pretty accurate too.

[Jolly]

Aren't there multiple types of AI? The kind that is capable of self-teaching and individual learning along with the type that just basically copies the behaviors of humans. I feel it's kind of weird that humans, who can't even understand themselves, much less each-other are projecting their mindsets and fears upon a machine that does not even exist yet instead of actually putting an coordinated effort into achieving a solution.

Personally speaking, if we could just get an intelligent AI capable of only solo-solving the laws of physics, math and computing for the benefit of mankind then that would already be more than necessary, I doubt there is really a need for "humanoid AI" and at the current moment, as things have always gone, all scientific breakthrough is military first, society second.

Basically speaking, we have no way of predicting what the fuck will happen when we finally succeed in the creation of an AI no matter what kind it is but I don't think scientific progress should halt because of some meager fear people have. Look at nuclear physics, sure, we might all die from nuclear war but when used righteously, it is a source of clean energy world-wide! (Let's ignore the radioactive waste, mining and warmongering for a moment)

[Boudica]

Aren't there multiple types of AI? The kind that is capable of self-teaching and individual learning along with the type that just basically copies the behaviors of humans. I feel it's kind of weird that humans, who can't even understand themselves, much less each-other are projecting their mindsets and fears upon a machine that does not even exist yet instead of actually putting an coordinated effort into achieving a solution.

Personally speaking, if we could just get an intelligent AI capable of only solo-solving the laws of physics, math and computing for the benefit of mankind then that would already be more than necessary, I doubt there is really a need for "humanoid AI" and at the current moment, as things have always gone, all scientific breakthrough is military first, society second.

Basically speaking, we have no way of predicting what the fuck will happen when we finally succeed in the creation of an AI no matter what kind it is but I don't think scientific progress should halt because of some meager fear people have. Look at nuclear physics, sure, we might all die from nuclear war but when used righteously, it is a source of clean energy world-wide! (Let's ignore the radioactive waste, mining and warmongering for a moment)

We already have several types of AI in our daily lives - Google search being a major one. I think creating humanoid AI is part of the human obsession to work out what we are and how we exist as we do. It makes sense. If you keep making a product a keep changing one thing at a time, you could see what factor caused the product to change from a robot into "intelligent life". It's all very far fetched but extremely fascinating.

I do very much agree with you, it's not something we need as a race. It's simply something we are fascinated by. I'll bet that military budget is much bigger than the social one too!

I also agree that there's no way to know what will happen, although, I enjoy the guessing game.

Nuclear energy is already becoming bigger in the UK and beginning to replace our fossil fuel power plants. Sure, the waste etc is questionable, but it's a step closer to that ideal clean energy many countries are striving for.

[Ion]

I think machines this smart will find loop holes around the fail safes. Just like in iRobot, with Will Smith.

It can only find loop holes around them if they are programmed to.

[Boudica]

It can only find loop holes around them if they are programmed to.

Could human error have a part to play or would it have to be via programming intentionally?

[Ion]

Could human error have a part to play or would it have to be via programming intentionally?

It wouldn't make sense if it wasn't programmed intentionally, unless the artificial intelligence could program itself (which would also be intentional as you would have to code something to make it do that).

[roger_smith]

It wouldn't make sense if it wasn't programmed intentionally, unless the artificial intelligence could program itself (which would also be intentional as you would have to code something to make it do that).

because all the vulnerabilities that are found in operating systems, CMS, SCADA systems, etc are all intentionally placed there. There could be a vulnerability in the code of the AI that allows an attacker to bypass controls. The interesting situation is when the "attacker" is actually the AI itself....

As far as AI programming itself being intentional, the whole discussion is about learning AI. This concept is not radical at all. Learning algorithms exist in the real world today with cars for example. Many transmissions have a learning phase they go through to adjust to how the operator drives, customizing shift points based on the driver's input.

[w00t]

Machine learning is already nearly omnipresent, I don't think I've used a piece of technology from 2010 onward that didn't at least have a small component of machine learning. Perhaps the most visible example is how Siri( and other similar services ) get better at understanding the user through continual use.

[Ion]

because all the vulnerabilities that are found in operating systems, CMS, SCADA systems, etc are all intentionally placed there. There could be a vulnerability in the code of the AI that allows an attacker to bypass controls. The interesting situation is when the "attacker" is actually the AI itself....

As far as AI programming itself being intentional, the whole discussion is about learning AI. This concept is not radical at all. Learning algorithms exist in the real world today with cars for example. Many transmissions have a learning phase they go through to adjust to how the operator drives, customizing shift points based on the driver's input.

What I meant was that for the AI to exploit itself and use it for mayhem, it'd need to be programmed to do so.

[troubler]

Stephen hawking rambles on about the dangers of AI (his field is physics so yeah, while he is intelligent not really his domain so to speak)

Then you have got Ex Machina, That one with Johnny Depp (can't remember what it's called but it's recent) you've also got Hal and of course most recently you have got Halo 5 which has always featured bat shit insane AI.

Yeah, Fuck Cortana right now. Seriously.