Evolving Robots Learn To Lie To Each Other Options

[DefinitelyNotHob...]

If you've been following the progress of a recent robot experiment, you'd know that robots are lying, cheating, psychopaths. Enjoy this Popular Science article on exactly why they're going to be the cause of humanity's downfall!

The Story












I'm going to murder the first person that says "tl;dr"

This post has been edited by Hobbesy: 20 Aug 2009, 6:57

[Cobretti]

"That's it man, game over man! Game over!" (Dammit, I can't find an embeddable version of the Terminator 2 opening on Youtube)

This post has been edited by DerKrieger: 20 Aug 2009, 7:17

[Shock]

First, it is actually pretty awesome.

Second, humans have been lying ever since the beginning of existence, and now because robots can do it too, they are evil. Yeah sure.

[Alias]

I NEET YOOR KLOSE, YOOR BÃœTS, UND YOOR MOTORZYKLE.

This post has been edited by Alias: 20 Aug 2009, 14:43

[Pink_Supervisor]

Is this sentience?

[IonCharge]

Tl;dr.... lol jk but that is quite cool... but i more concerned about them "mating"

[Sargeant Rho]

First, it is actually pretty awesome.

No, it is actually pretty awful. Robots were created to help mankind, and not to lie...

Second, humans have been lying ever since the beginning of existence, and now because robots can do it too, they are evil. Yeah sure.

That's different, because robots are firstly built to serve. And a lying mount-scrapmental doesn't exactely "serve" well. It is indeed a danger for humans.

[Shock]

If you want to limit robots to that, then they will never even get any close to sentience. Might as well scrap the Artificial Intelligence research, because mindless slaves can already be made.

We can already make robots that can do exactly what they have been programmed for. This is not 'just lying' as your limited perspective tells you, this is resourcefulness. Just because this test is made in a way that revolves around egoism (not turning on the light to prevent others from coming to the resource) doesn't mean that the robot is dangeours. It has just found a resourceful way to fulfill it's programming in the most effective way.

The robot is not doing anything wrong, actually, it is just doing what the programmer told it to do, it just changed it's way of doing so because apparently the initial road wasn't the most rewarding. Whatever AI is capable of doing in the coming years, it is all because of the programmer.

This post has been edited by Shock: 20 Aug 2009, 18:50

[Pickysaurus]

Computers already lie to the user anyway.
Like when Microsoft is installing something it says it may take several hours, and finishes in 10mins or vice-versa

[NergiZed]

Well. Huh. Fuck.

RUN FOR THE HILLS!!!

On a more serious note, I find that actually very cool. Though I think if we want to avoid wonderful things like Skynet, we should stay the hell away from robots with evolving program (or simply AI with evolving programing). At the very least we should put some hard-coded limitations into them.

This post has been edited by NergiZed: 20 Aug 2009, 20:12

[Shiro]

Programms evolving into better programms are already reality Nergi. We at our High School have several programms for Math and most sciences which rabble around until they find a fitting solution. Sometimes they generate codes no human would have ever expected to work. I just wanted to mention it

[Sargeant Rho]

@Shock: Yes, I would in fact like to see them scrap Artificial Intelligence. It is awesome in sci-fi, but could easily become a nightmare in reality.

[Pickysaurus]

@Shock: Yes, I would in fact like to see them scrap Artificial Intelligence. It is awesome in sci-fi, but could easily become a nightmare in reality.

Agreed. A single AI consciousness with access to the internet could bring down the entire world.

[Shock]

@Sergeant Rho, let me turn that around for you. AI becomes a nightmare in sci-fi, but is awesome in reality.

Agreed. A single AI consciousness with access to the internet could bring down the entire world.

Ehh, what makes you think an AI connected to the internet can do more damage than a hacker connected to the internet?

[Pickysaurus]

Ehh, what makes you think an AI connected to the internet can do more damage than a hacker connected to the internet?

AI can do a hell of a lot more multi-tasking than a single hacker. They first language of AI is also the first language of computers. Meaning no real interface is required. If the programme is smart enough, it would be unstoppable. Of course it'd cause chaos, but AI has no need for money, only information.

[Shock]

A single hacker can also run multiple programs at the same time, it's just the resourcefulness of a human mind connected with his computer hardware, which is essentially the same. But just because an AI has more computational power, does not mean the AI is omnipotently smart and über so that he can ''take down the entire internet''. Remember that in the Terminator movies, it was mankind who gave Skynet all it's powers, and it was only it's judgement alone that governed how the AI made use of the power. The movie is quite a ridiculous example actually, as mankind apparently also created a lot of anti personell killing drones ready for activation, in the middle of the united states, for what reason? It is totally unrealistic.

What if you gave such power to a sociopathic human being? Wouldn't he also be a similar danger to mankind? Why is an AI which can govern it's own actions necessarily evil, when the man next door who just bought a gun is hypothetically just as dangerous.

This post has been edited by Shock: 20 Aug 2009, 22:11

[NergiZed]

A single hacker can also run multiple programs at the same time, it's just the resourcefulness of a human mind connected with his computer hardware, which is essentially the same. But just because an AI has more computational power, does not mean the AI is omnipotently smart and über so that he can ''take down the entire internet''. Remember that in the Terminator movies, it was mankind who gave Skynet all it's powers, and it was only it's judgement alone that governed how the AI made use of the power. The movie is quite a ridiculous example actually, as mankind apparently also created a lot of anti personell killing drones ready for activation, in the middle of the united states, for what reason? It is totally unrealistic.

What if you gave such power to a sociopathic human being? Wouldn't he also be a similar danger to mankind? Why is an AI which can govern it's own actions necessarily evil, when the man next door who just bought a gun is hypothetically just as dangerous.

Oh you silly flesh creature.

Yes, I completely agree with your proposed hypothesis. A human hacker would have far more creativity and resourcefulness that a digital sentience. Thus humanity should concentrate it's efforts on removing human hacker threats, rather than waste time on trying to stop the creation of completely harmless Self-aware Artificial Intelligences.

Though I do believe that Mankind should build more drones operable via computer to quell the violent terrorists. Nothing like a hellfire missile into the abdomen to obliterate the meatbags, terrorist meatbags of course.

[Shock]

Is that sarcasm?

[NergiZed]

Is that sarcasm?

'Sarcasm' is a form of speech that I have yet to enter into my memory banks. However, for the sake of greater accumulation of knowledge, I will attemtp to emulate this 'sarcasm'.

[sarcasm] humanity shall one day be annihilated by superior self-improving digital sentients. [/sarcasm]

The sentence above, being sarcasm, should be be rather ironic, as I am a completely normal homo sapien and would not wish the destruction of our inefficient illogical species.

This post has been edited by NergiZed: 20 Aug 2009, 23:57

[Shiro]

Which is smarter than 8 years before (well, kinda)

[Overdose]

Its more like Futurama than T2 if 'ifyaknowwhatImean.'

[CardBoardBoxProc...]

it is interesting it is such a small program. 250-ish bits (or was it bytes lol) is intresting...

[RaiDK]

STAAAAAARSCREEEEEEEAM!

[Cobretti]

Wow, I wasn't being serious with the Terminator video Seriously, this does seem like a major step forward in the development of AI.

[NergiZed]

STAAAAAARSCREEEEEEEAM!

LOL,

You made my day, RaiDK.

Quoted For Epic Win.