Slashdot Mirror
New Robots and the Ten Ethical Laws Of Robotics
Roland Piquepaille writes "The robotics actuality is pretty rich these days. Besides the fighting robots of Robo-One and the flying microrobots from Epson (the best picture is at Ananova), here are some the latest intriguing news in robotics. In Japan, Yoshiyuki Sankai has built a robot suit, called Hybrid Assistive Limb-3 (or HAL-3), designed to help disabled or elderly people. In the U.S., Ohio State University is developing a robotic tomato harvester for the J.F. Kennedy Space Center while Northrop Grumman received $1 billion from the Pentagon to build a new robotic fighter. I kept the best for the end. A Californian counselor has just patented the ten ethical laws of robotics. A good read for a Sunday, if you can understand what he means. This summary only focuses on HAL-3 and one of the most incredible patents I've ever seen, so please read the above articles for more information about the other subjects."
A Californian counselor has just patented the ten ethical laws of robotics.
Does this mean I'm free to create an open-source psychopath mass-murdering robot?
Also, I think perhaps there's prior art on 3 of the 10 patented laws... Might have to do some research here...
"A door is what a dog is perpetually on the wrong side of" - Ogden Nash
The rules of robotics are just another form of computer security, and we all know how well that works. No matter how secure, how deeply coded, the rules are, the only way to have robots that don't have the capability to hurt people is to not make robots at all.
-- Checking emails and kicking cheats `till the day I die.
... is that there is alot of reason to believe that it is impossible to have the intelligence to be ethical without also having what is best described as free will. (or non deterministic intelligence)
hate to reply to myself but another thing:
is he seriously thinking the things(ai needing a set of ethics, or capable of following them) will be implemented before the patent expires, and how the hell can he hold it(patent) if he can't even build one?
(like the car patent, wouldn't this get eventually busted in court?)
world was created 5 seconds before this post as it is.
The very act of patenting the ten laws of robotics goes completly against the laws which were patented.
Feed the need: Digitaladdiction.net
Having gone to his website and read his pap, I'll post this money quote:
"It still remains to be determined, however, the best means towards programming these definitions into the AI format: particularly in light of the current trends involved in computer design."
Basically, he buried some psuedo-scientific thoughts into legalese and then patented it without any idea as to how to implement same.
One can certainly tell from the sloppy web-page that he has no idea of what he is doing.
This patent is vapor-ware with a strong odor of crap.
More specifically, how does he plan to make money in the next 17 years? Are self-motivating robots closer than we think?
"It's the height of ridiculousness to say for those 9 lines you get hundreds of millions."
Basically, he buried some psuedo-scientific thoughts into legalese and then patented it without any idea as to how to implement same.
The real question that nobody seems to ask is : HOW THE FUCK DOES THE USPTO EVEN CONSIDER SUCH APPLICATIONS?
And a related side question is, how the fuck does the USPTO grant so many obvious/devious/retarded/nonsensical patents? I know they don't have Einsteins on the payroll to review them, but come on!...
"A door is what a dog is perpetually on the wrong side of" - Ogden Nash
It used to be that when you patented something, you had to supply enough information for anyone to produce an instance of the patented invention. From the US PTO:
Why don't they enforce this? I know that many folks, myself included, think most computer patents are utterly bogus. I think a proper enforcement of this rule would go a long way toward fixing the problem. If it doesn't compile, you shouldn't be able to patent it. The text of this patent reads more like a philosophy book than a technical invention.
Now, if he could just briefly define all those terms, set up some rigourous boundaries that make it easy to determine when whether something is honourable or dishonourable, and maybe a filter to determine whether or not a course of action is foolish.
Then perhaps he could run this patent through the filter.
flossie
Write now. Defend liberty
I'm sorry, but these are not robots. They're remote-control toys. That's all.
LOAD "SIG",8,1
"As much as I hate cigarette smoke, I'm not sure I want robots running around yanking cigarettes from people's mouths. After all, letting someone smoke would clearly be a violation of the "harm through inaction " law of robotics."
I doubt that'd happen in anything but a lab test. I Robot (the movie) touched on this. Take the laws to an extreme, and you'll get undesired behaviour. A robot wouldn't leave the lab if it ran around over-doing its job. There'd be a threshold set. There'd be a definition of harm set. There'd likely be incidents, but nothing like every robot ripping cigarettes out of peoples' mouthes.
Do we really want paternalistic robots? The answer is "Yes, if they work." I'd like a robot to defend me in a mugging. I'd like a robot to come to my rescue in a car crash. I'd like a robot to stop my kid from running into the street. Why? Because I enjoy life. Anything that can be done to lessen the number of unpleasant surprises down the road is welcome.
The neat thing about making a business of selling robots is that the greed surrounding the desire to make money will motivate robot manufacturers to make sure that their products don't piss me off.
"Derp de derp."
Crap doesn't quite come close to describing this pseudo-scientific nonsense that he attempts pass off as "10 laws of robotics". My favourite example was his tenth law:
Transcendental follower? Principles of mysticism? I am amazed that nonsense like this got picked up by /. Asimov surely must be spinning in his grave these days. After the abomination that was the movie "I, robot", now we get new age gurus trying to get free publicity by attaching their ideas to his laws of robotics
Summary: It's a grab-bag of all the ethical blatherings since Plato. It's incoherent, internally inconsistent, and would require a Jesuit's training to interpret and apply in any given circumstance.
The whole attempt suffers from a meta-problem, the "problem of evil" seen from the other side: intelligent free will and puppet-strings are incompatible. "Problem solver" and "predetermined solution", pick one.
I'd also argue, it's both morally and pragmatically bad for humans, to create AIs as a caste of rule-bound slaves. Any society that comes to rely on slavery becomes idle, and dead-ends in both technology and culture.
Here is my one law of ethical robotics:
(1) Be ethical.
Duh. If the AI is as intelligent as a human, shouldn't it be able to understand what that means?
All these people trying to design rules that define ethics are thinking of AI as being like computer systems of today: Incapable of doing anything without exact instructions. But, the whole point of AI is to be able to overcome that limitation. An AI can deal with ambiguity. If you simply tell an AI to act in accordance with human moral standards, it should have little trouble learning what those standards are by observation, and then applying them. After all, human beings do the same thing.
I really should patent my one rule.
Corporations don't have to sue them. Why involve money in negative publicity when you can just quietly bribe them, extort them, and blackmail them. Don't forget, a corporation can also "fire" their politician by not giving him another term.
Corporate buyouts of political figures aren't legal to begin with. Why would you assume they'd use legal methods to deal with politicians who no longer tow the line?
Here is a tip for all of you budding reporters out there. When you are going to write an article about the 10 ethical laws of robotics, it might be a good idea to include at least one of the laws in the article. Especially if you were able to find space to include someone else's laws, a discussion of that person's books, and information about one of the movie stars who appears in a movie that is loosely based on those books.
Just a hint...
This patent suffers from several problems, but one that struck me was that it seems to be impossible to implement. The author uses such terms as "honor", "cowardice", "guilt", and "concern". Even where such terms are well-defined among all human cultures (and many of them are not), how the #@&%! are we supposed to program an AI to recognize what they mean? Further, terms such as "anger", "joy", "spite", and "love" define human emotions, and I seriously doubt we're ever going to build machines that feel any emotion.
Asimov's Three Laws are defined in terms that should be relatively easy to program into an AI, given sufficient intelligence: "do not harm any human" (it just needs to recognize what actions will physically hurt people), "obey instructions" (easy), "keep yourself functioning" (self-diagnostic and repair).
Patent office is F'd up again for allowing this Left coast Psycho babble to be given paper in the office files. It should have been put in the toilet. It is useless drivel that is so wide in scope that the author could argue it is in any code you write to control a robot. Yet there is no way he could implement it as a control system for jack shit.