Have we Broken the 3 Laws of robotics?

[StormShaun]

To be Frank (Not West lol) I just want a AI or VI like EDI form Mass Effect 2 or a danm cybotic parts such as arms, legs and etc because that can help disabled people and mabey a nano suit from Crysis....that would be awesome

 

[DefunctTheory]

You are under the assumption that AI exist.

You are wrong.

AI does exist, it's just extremely early stages atm.

Spoiler

OT: I think the 3 laws only really concerns AI, and the weapons of today are simply machines and good computing.

In a technical sense, you are correct. Thats AI.

The Rules of Robotics, however, do not refer to that definition.

What about Watson?

http://www.scifi-review.net/ibms-watson-beats-jeopardy-champions-ken-jennings-and-brad-rutter.html

He interacts with people without being manipulated and can freely answer questions. Quite well too.

I'd argue it's probably able to learn in a tradition sense. In what way is that not AI.

Watson is a computer program/server with no real world observational skills that used wikipedia to win Jeopardy.

Its basically a statistics engine.

I hardly call this AI. Its just clever programming.

 

[WittyInfidel]

You are under the assumption that AI exist.

You are wrong.

AI does exist, it's just extremely early stages atm.

Spoiler

OT: I think the 3 laws only really concerns AI, and the weapons of today are simply machines and good computing.

In a technical sense, you are correct. Thats AI.

The Rules of Robotics, however, do not refer to that definition.

What about Watson?

http://www.scifi-review.net/ibms-watson-beats-jeopardy-champions-ken-jennings-and-brad-rutter.html

He interacts with people without being manipulated and can freely answer questions. Quite well too.

I'd argue it's probably able to learn in a tradition sense. In what way is that not AI.

Watson is a computer program/server with no real world observational skills that used wikipedia to win Jeopardy.

Its basically a statistics engine.

I hardly call this AI. Its just clever programming.

It was disconnected from the internet.

It only used whatever it already knew, by learning. Presumably by surfing the net previously. But there's no requirements of AI that says where you must learn information. Just like there's nothing that states AI has to have real world applications. Just like real intelligence doesn't.

Out of curiosity. What exactly would constitute AI for you?

For it to be true AI, it would have to be self-aware. "I think, therefore I am" kinda stuff. Otherwise, it's not true AI. It would simply be an advanced program, no more intelligent than say...a washing machine. It would have to have the capability to learn, enact what it has learned, and work in the abstract known as imagination. Being able to imagine, I think, would be key.

 

[DefunctTheory]

You are under the assumption that AI exist.

You are wrong.

AI does exist, it's just extremely early stages atm.

Spoiler

OT: I think the 3 laws only really concerns AI, and the weapons of today are simply machines and good computing.

In a technical sense, you are correct. Thats AI.

The Rules of Robotics, however, do not refer to that definition.

What about Watson?

http://www.scifi-review.net/ibms-watson-beats-jeopardy-champions-ken-jennings-and-brad-rutter.html

He interacts with people without being manipulated and can freely answer questions. Quite well too.

I'd argue it's probably able to learn in a tradition sense. In what way is that not AI.

Watson is a computer program/server with no real world observational skills that used wikipedia to win Jeopardy.

Its basically a statistics engine.

I hardly call this AI. Its just clever programming.

It was disconnected from the internet.

It only used whatever it already knew, by learning. Presumably by surfing the net previously. But there's no requirements of AI that says where you must learn information. Just like there's nothing that states AI has to have real world applications. Just like real intelligence doesn't.

Out of curiosity. What exactly would constitute AI for you?

It was disconnected from the internet, but had the entirety of Wikipedia stored in its memory, along with various other sources of information. It didn't learn squat, it was just loaded with the appropriate data and given a search function. It then found appropriate answers, gauged them with statistical algorithms, and chose the most likely answer.

Basically, it googled its own hard drive.

In my opinion, to have 'true' AI (Distinguished from the actual field of AI, which has a pathetic expectation of itself), a computer has to be able to actually conceive its own methods of analysis and perception, based on a standard programmed into it in advance. In both examples you provided, the subjects were merely exercising algorithms and coming up with solutions entirely constrained by its programming.

In other words, current AI basically just crunch numbers. 'True' AI should be able to formulate entirely new algorithms.

 

[wolf thing]

i was under the impression that we used theses laws for real robotics, useing them after the book but from what people are saying this is wrong. ah well. we probably should there not bad laws we just need to make sure we dont have a incredibly intelligent robot leading them because that never gos right

 

[Chris646]

No, I do not believe that we have broken the laws of robotics. The computer may send instructions to a missile, but it is still the missile hitting the targets. Enemy AI is not hurting real human beings, just virtual ones that respawn. If the gun is pre-programmed to auto-shoot at humans when they are seen, that breaks the laws.

 

[WanderingFool]

You know the 3 laws are fictional right? From I-Robot?

Robots are gonna do whatever the hell we want them to, and seeing as we're human, killing is pretty high on the list.

Way before I-Robot, but since I-Robot was suppose to be an adaptation of one of Asimov's books...

Anyways, I seriously doubt it.

 

[Glademaster_v1legacy]

Well that is more for AI as people have already said. At most we have only made advanced programs or come close to VIs but not AIs yet.

 

[DarkRyter]

Their AI isn't complex enough to register as Asimov robots.

 

[mitchell271]

I don't think we have. The 3 laws apply to AI controlled robots that are designed for everyday use. At least, I think so. If it applies to all, soon we will. If it applies to the everyday, no, not yet.

 

[DefunctTheory]

What you explained it doing is essentially what we do.

How do you think general knowledge works for humans?

They store information from various sources and access it when we need to.

Our method of recall may be more convoluted and our storage capacity is not as refined. But we essentially do the same thing.

The only difference between a human and Watson in terms of intellect, would be level of self-awareness.

If you were to combine something like that with complex image recognition, like Natal, and slightly more advanced learning software.

Like the robot in the video that got posted to you earlier. The Honda robot.

Than I'd argue you'd have AI in the truest sense of the word. Or atleast a budding form of it.

I don't think Watson is the pinnacle or even as you put it, true AI. But it's certainly a milestone for it.

It's not just able to retain information. It can recall it for situational use when it's interacted with.

That's certainly a step in the right direction.

True AI, as I put it (Again, I am not referring to the science of AI, as their definitions are looser then a 5 dollar hooker), will be able to create knew was to interpret data.

Current 'AI' cannot do this.

And no, from my perspective (Again, this is my opinion), these devices are NOT a step towards AI, as AI should be able to redefine its own parameters, not just regurgitate numbers based on algorithms not of its own design (The Human mind does both, by the way).

 

[Dwarfman]

You know the 3 laws are fictional right? From I-Robot?

Robots are gonna do whatever the hell we want them to, and seeing as we're human, killing is pretty high on the list.

Issac Asimov's three laws were indeed meant for his sci-fi novels. That does not mean they should be dismissed. They are in fact very simple and prudent guidlines for anyone willing to dabble in AI (Especially morons who indeed want to use the tech for violence)

I have been wondering on this for a While.

we have made missiles that are Computer operated to hit a target

Built Guns that work with cameras.

so have we already Over stepped the rule. We have Programmers working all the time making AI's to kill the players of games and adapting tactics. (if you piss of your AI you are fighting against and it fights back by hitting the real you with a missile)


These are the 3 Laws.

1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.
2. A robot must obey any orders given to it by human beings, except where such orders would conflict with the First Law.
3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.


I Would like to see your reasoning to this.

I think you missunderstand the difference between a computer programme and a robot or AI. Both of the weapons you describe are in the end opperated by a human being and are incapable of independant thought. The AIs in a computer game only act the way they do so as to simulate a 'real life' entity. These programs should not be called AI, instead they should be called CS (computer simulation).

 

[Wintermoot]

it only applies to real robots (for example a combat drone violates the first rule)
and yes they are already broken, for example there was this guy in Eastern Europe that got attacked by a factory robot (even though it was programmed to do so by a human)

 

[Thaluikhain]

People find it hard to understand, or even define, "intelligence".

It's not surprising that creating our own version isn't going to happen for a bit.

 

[ParkourMcGhee]

who is to say there is no ghost in the machine to come in the near future

The neurosurgeons who freely admit they have no freaking idea what half our brains are for.

That's who can say.

That's not true!

We know what more than half our brains are FOR... we just don't exactly know HOW they do it!

source: Doing biomedical engineering at uni, had 10 neuroscience lectures so far.

For more information google Purkinje cells cerebellum, or stuff about retinotopy - we have a vague idea, but no proven path how stuff like this happens: http://www.dailymail.co.uk/sciencetech/article-1081303/Scientists-discover-single-brain-cell-dedicated-Jennifer-Aniston-inside-womans-head.html

EDIT:

OT: I actually want to make a one of those nerf turrets with a webcam from thinkgeek automatically shoot anything that moves - I'm doing image processing next year at uni.

So if those laws apply to any machines, then yeah, we did that ages ago.

If they only apply to AI however, go talk to one of those online robots and ask them if they're hooked up to a nuclear missile... Have fun ^-^.