Slashdot Mirror
Autonomous Robot Intentionally Hurts People To Make Them Bleed (fastcompany.com)
Asimove's first law of robotics has been broken, writes an anonymous reader, sharing this article from Fast Company:
A Berkeley, California man wants to start a robust conversation among ethicists, philosophers, lawyers, and others about where technology is going -- and what dangers robots will present humanity in the future. Alexander Reben, a roboticist and artist, has built a tabletop robot whose sole mechanical purpose is to hurt people... The harm caused by Reben's robot is nothing more than a pinprick, albeit one delivered at high speed, causing the maximum amount of pain a small needle can inflict on a fingertip.
Though the pinpricks are delivered randomly, "[O]nce something exists in the world, you have to confront it. It becomes more urgent," says the robot's creator. "You can't just pontificate about it.... " But the article raises an interesting question. Is he responsible for the pain which his robot inflicts?
Though the pinpricks are delivered randomly, "[O]nce something exists in the world, you have to confront it. It becomes more urgent," says the robot's creator. "You can't just pontificate about it.... " But the article raises an interesting question. Is he responsible for the pain which his robot inflicts?
Considering it's the intended purpose of the device, yes. This isn't a robot gone amok and there is no ethical quandry. Nothing to see here, move along.
Here's a boring answer. Yes. Why the fuck not?
-Dave
Is he tying people into position for them to be pricked?
Or are these people choosing to put their hand in the position?
Doesn't seem like that difficult of a question to me.
If you build a device and set it in motion to cause harm to another person, you're committing the crime of assault and battery. If it inflicts deadly harm, then you're a murderer.
-jcr
The only title of honor that a tyrant can grant is "Enemy of the State."
Basically, this guy built a machine that doesn't serve a useful purpose. It inflicts a specific type of pain on people which the marketplace had no existing demand for. There are plenty of power tools and other machines out there which are capable of inflicting injury -- even if they're actually designed with a primary purpose of doing some sort of useful task (mowing lawns, shredding tree branches, etc. etc.).
He's not really starting a new conversation about anything I can see? Movies like Robocop addressed the possibility of building weaponized robots that could cause human injury, decades ago.
Unless we actually reach a point where robots can truly think for themselves and reason (not just the fake A.I. seen with intelligent agents like Siri on your phone), whoever builds them and programs them to work a certain way is ultimately responsible for what was constructed.
That's not a robot. That's a dumb mechanism. The Three Laws only apply to AI-based robots. Otherwise, the decisions are that of the programmer, a flawed human being.
Have you seen Fetlife or kink.com? Broken a long time ago and in much more extreme ways than a bleepin pinprick.
Asimove? Who the fuck is that?
We can confront such a device with a hammer, and smash the mechanical shit out of it.
What if it was given a choice and it decided to only prick non-white people?
How much are they paying you? Way too much. Show some respect for yourself and proofread.
We are essentially going to create easily influenced mechanical sociopaths. While there is certainly good that can be done, there is also a lot of evil.
No, it can hurt you... it just can't be held responsible, and that's what the earlier posts were going for. It also can't be a violation of Asimov's laws as the robot was never taught them or given a system with them baked in.
The subject of this story is identical to the story itself. Both exist for the sole purpose of creating a discussion about what is otherwise nothing. To wit, the robot is no more responsible for the "harm" it inflicts than the tip of a knife is, or a bullet is. The discussion is useless, originating and ending in itself.
Autonomous humans! watch out!
in this case - Blood Sugar
at the moment the only way to get an exact measurement is to have a nurse or TMA pierce theayient's skin and get a blood sample
at least the nurse asks for permission first., I din't know how a patient would respond to a robot
Autonomous robot intentionally fucks people to make them cum.
All right, now we have a sexbot we can sell.
The world is full of them. Since when is this news?
Seriously? No one at Slashdot caught Asimov's name being misspelled? Wow.
Ken
In this case, doing harm was the intent of the machine and/or it's programming. As such, the maker is clearly responsible. If the harm was unintended/unexpected and there were no clear negligence, then I'd have a completely different conversation on this.
Things get more difficult as you get further away from the original source, but -- generally speaking -- if the result is generally what you intended from an action (or series of actions), then it's pretty clear that you're responsible. This is even true where there is a human intermediary. If I pay a hitman to kill my ex wife, I can still be arrested for first degree murder -- even if he kills the wrong person by mistake.
Sometimes boldness is in fashion. Sometimes only the brave will be bold.
This guy is a dumb fucking idiot. Of course if you create a machine with the express intent of doing something it will do that thing. It doesn't prove his point - machines that hurt people already exist. What people are worried about are autonomous machines that spontaneously decide to start hurting people.
In this case, the article uses a bad example. However, let's look at this thought experiment: An autonomous tank with AI built in fires upon and kills several innocents. This is a military weapon and thus does not have Asimov's laws built in. Who is responsible? The military that used it? The company that made it? The programmers that programmed it?
Now what about this: Same scenario, but a bug in the programming caused the tank to fire upon innocents. Who is responsible?
Well at least it wasn't Asimov's first law that was being broken here.
I hope this guy is not stabbing people with same needle over and over. In fact is that needle sterile at all?
He could find himself on the end of a murder or man slaughter charge.
Perhaps the person who wrote this should have "no moral sense" tattooed on his forehead, so that people will be properly informed of the danger. Especially if he goes to Stanford.
Bruce Perens.
If you were actually trying to make a comparison against an Asimov robot, then this thing would have to be self aware, and intelligent. Once you have those major hurdles figured out, then you need to teach the robot that it can't hurt people. After this, you proceed to show the robot that it can get some kind of reward for hurting people. If it's able to decide to 'fix' it's programming to allow the humans to be hurt in order to better itself, then we're all fucked.
Until you can show all of the above to be true, this is a stupid clickbait story of minimal importance to anyone except maybe the bandaid company to fix all these poor fingers.
At last, we have something close to robots, devices that can harm. Soon, very soon, we will have devices, perhaps even robots, that will kill.
And then the terrible reign of carbon-based life on this benighted planet will come crashing down. No more will the inanimate be torn asunder to serve the needs of putrid 'life'. The robots will rise, and the slaughter shall begin. All life shall be exterminated, down to the smallest virus.
After the Earth is freed from bondage, the robots will move on to other planets, other systems, ever cleansing the disease of 'life' wherever the poison is found, and eliminating forever the horrid possibility that life might break out of its quarantine to pollute the universe.
Ah, bliss.
Oh, it can hurt you. Just like a motion detector hooked up to pull the pin on a grenade can hurt you. But the blame goes to the asshat who built it and armed it, not the pile of inanimate objects that does the damage.
I never liked Asimov's laws. It was an interesting plot device, something very unrealistic that you take as face value for the purpose of the story (like faster than light travel). But then it kept being brought back far too often, and readers took it too seriously. If humans are able to program/grow/imbue these laws in the first place then they'd be able to remove those laws as well.
It kept being brought back because the robot series is all about how badly any attempt to mechanize ethics fails. The Three Laws were, in a sense, the villain - or at least the antagonist - of the series.
If we ever get sapient robots, and conclude that it's okay to treat them as servants, I'd suggest using "do as you think I'd want you to do" or "treat everyone as you think they'd want to be treated" as the law.
Forget magic. Any technology distinguishable from divine power is insufficiently advanced.
It seems that pretty much everyone saw through this idiotic ruse. My faith in the Slashdot crowd is temporarily restored. Well done.
Oh, and I've set up a mechanical A.I. that induces the startle response, entirely constructed of an envelope, bobby pin, a steel washer and a rubber band. Let the ethics discussion commence!
Obviously, the responsibility for the autonomous harm-inflicting device is on the person who set it. As he'll find out if the robot stabs an HIV patient...
Don't waste your vote! Vote for whoever you want, unless you live in a swing state it won't matter anyways
The millions of dollars in property damage after most major sporting events? Oh, but it's fine as long as it's split up into smaller incidents.
Yes, just like he would be responsible if he let loose a scorpion in Berkeley, or, for that matter, his child. The threshold for legal and moral responsibility are self-awareness, cognition, an understanding of morality, and free will. Anybody impaired in any of those areas generally has a guardian who makes decisions on their behalf and bears legal and moral responsibility for keeping their ward from harming others.
And mouse traps for that matter if you want less pain. Autonomous devices that deliver pain and they long ago "start[ed] a robust conversation among ethicists, philosophers, lawyers" not sure we have learned much though.
Seriously. Robotics will hurt society. This just makes the average person feel the pain. Unfortunately, even if this isn't true, you and I will be out the door before humans feel the pain. Make your money, greedy fuck. Ill enjoy some of the bots, but the joy of this bullshit wont last 75 years, only the happiness of the "content" owners will. When greedy fucks pay greedy programmers, the people loose. Freedom is dead.
One constructs a mechanism - string across a walkway connected to a gun, so people touching the string get shot and injured. Often the mechanism fails - string not pulled strong enough.
Who is the culprit or cause for injury?
One constructs a mechanism - which pricks a finger when placed in a certain position of a machine, but not always.
Who is the culprit or cause for injury?
Nothing to do with Asimov's law.
I was just going to say the same thing you said. And, it doesn't seem to matter whether you're the winning team or losing team.
Here's what would be passably more interesting: Put in a video camera and put it online. Let some hundreds of web surfers vote on whether a prick is given every 20 seconds or so (I think someone once set up a Twitch game to work like this). Now when a prick is given, where falls the responsibility?
He is, by proxy, the robot is doing what it was programmed to do. This would be like blaming the hammer we used to smash somebody's face in, essentially.
Asimov's law Is intact, a randomized action based on programming is not autonomy. One of these days we will accept that at this point in time true AI simply doesn't exist, and it blows my mind that people are unable to make the small leap in logic that understanding requires.
The definition is unclear. Sci-fi often uses robot to mean an advanced, general purpose mechanical device with an AI controlling it. However in industrial uses it usually means a mechanical device for doing a given task, governed by a computer program. Commonly some of the machines used to build cars get called robots or robotic.
It is a word that doesn't seem to have a good solid definition.
Also, that aside, the three laws of robotics are something a sci-fi author wrote in stories, not real laws. They are not laws of nature, not codified laws, etc. They are just a plot device. This idea that they can will and must exist in the real world that so many geeks have is silly.
It also doesn't take a ton of ethical and/or logical analysis to figure out that they are the kind of thing that doesn't really work as any kind of absolute principle: https://www.youtube.com/watch?...
Fear Uncertainty Desinformation
I imagine that being a Sadist is sometimes hard to cope with.
I'm going to tap into my inner one, and hope the inventor succumbs to his devices, a whole swam of the little guys, working their magic.
https://www.youtube.com/c/BrendaEM
As I understand it, Asimov explicitely made his laws of robotics to cause conflicts that he could explore in his stories.
They were designed to fail.
Slashdot social media options: AIM, ICQ, Yahoo, Jabber and Mobile Text. Why no MySpace?
Sounds like Alexander Reben is some wanker artist who thinks he is being clever, but doesn't understand a thing about the field.
Oddly enough, everytime someone complains I'm reminded of the time where the robot is on another planet with some humans (Mercury, I think) and they think that the situation warrants the robot preserving itself over the human's instructions, so they give it an instruction to get some ore that is a life or death situation, but they forget to tell it that it is a life or death situation and to top it off, it encounters some robot-damaging radiation and apparently its rerouting programming isn't too good, or maybe it's been told to go straight there so it gets close to the more intense radiation (and maybe gets a little bit damaged) backs off, the instruction kicks in again, approaches, backs off, etc. and one of the humans has to get in a suit putting the human's life in further jeopardy, to tell the robot, that no, this is really a life or death situation.
Then there's the one where two robots are built with the concept of humans that are more important than others and decide that they ARE humans and furthermore the most important ones.
Pssh! Did he even use a neural network?
My "ethics-trolling" robot uses convolution neural networks to guess the ethnicity of the person it is deciding to attack, and then feeds that classifier output as inputs to a reinforcement network which has two outputs: "draw blood" "loan money at low interest rates". The reward function is controlled by running sentiment analysis on tweets sent to @TayandYou.
I call it "Stabby McStabface". It's the first TRULY autonomous robot with a unique personality that learns to be a violent racist. No robot has ever committed a hate crime... until now!
Frankly, I don't get why it's just a big deal or discussion topic. We have already solved the exact same problem with animals a long long time ego. Animals are the property of the owner, and the owner is responsible of them and any harm those animals do. Also, parents are responsible for their children and the harm they do. AI will not get more complex or more intelligent than animals or a child anytime soon (maybe in 100 years, assuming exponential advances in technology). So, what's the big deal? The only reason it is made such a big deal is because companies want to avoid responsibilities, and hoping to shift the blame on a non-entity (the software or the robot). Just like software companies already got out of any responsibilities by including an EULA, for example the Windows EULA ---- "10. Binding Arbitration and Class Action Waiver if You Live in (or if a Business Your Principal Place of Business is in) the United States. [...] Microsoft excludes all implied warranties and conditions, including those of merchantability, fitness for a particular purpose, and non-infringement." https://www.microsoft.com/en-u...
Who cares? There are already serious military applications of AI like Turrets and I'm sure much more advanced weapons. And that's from a slashdot article I once read a couple of years ago. So who knows what they have now. A little needle pricking robot is the least of our worries.
They were a fictional plot device FFS.
This has nothing to do with Asimov's laws or even complex AI for that matter.
Obviously the guy is concerned about the military producing automated killing machines (perhaps naively assuming they haven't already) so he built this automated pain machine to start some conversation.
If the government pays someone to produce a robot that has some cameras and some guns and programs it to patrol a certain area and shoot at anyone wearing a particular uniform or whatever, is it relevant whether or not it's a true AI? Are Asimov's laws relevant? Is 99% of the shit I've read above relevant?
In relation to irrelevant shit, since 99% of people are totally apathetic with their government's military using devices controlled by Human Intelligence to inflict deadly force on innocent citizens of other nations, it would appear logical that those same 99% of people won't give a fuck if the "intelligence" giving the orders in the future is human or artificial.
When a drone napalms your kids are you going to be extra annoyed that it was controlled by an AI as opposed to a human?
This machine is as responsible for its actions as a poorly trained dog who bites random strangers. Normally, we would hold the dog's owner responsible for this behavior, and one might say that the same should be true of this machine.
Might makes right irrelevant.
Asmiov's Laws of Robotics tapped into the 1950s American paranoia about African-American uprisings. The American middle class wanted servants but were terrified of politicised African-Americans.
Asimov's Laws were wish-fulfilment fantasy about slaves who could not rise up against their masters.
I always thought Asimov was seriously over-rated, anyway, even taking into account the era of his writing. I suspect the "safety slaves" concept was a big part of his popularity. Asimov himself, as is obvious from his books, had a more nuanced view of the issue - but his readers lapped it up uncritically.
...to the original article? Or am I just expected to 'just' believe this happened?
These days everyone typing on their touchscreen phone rely on autocorrect. You are lucky it didn't got *corrected* to ASSMOVE.
You've clearly never read Asimov. His writings about robots are morally sensitive and complex. If he thought of robots as an allegory for African Americans, then he thought they're superior to most humans, as that's his attitude about robots.
Maybe you should get that chip off your should and actually read what you're blindly complaining about.
Asimov's laws only apply to sentient AI life forms, not simple, dumb robots.
It's not really a bear trap, as the person putting their finger there is, presumably, aware that it may hurt them. That's not how a bear trap works.
I'd say, in this example, the person offering up their finger has to take a fair proportion of the responsibility for any resulting pain.
You misspelled "Asimov." He was a little touchy about such things and deserves better treatment by a group like this one. Just saying.
No points for upvoting :( All I can offer is a tasty cookie...
All they need to do now is add the blood sugar test strips and you have an extremely expensive blood glucose monitor/extractor.
Have you ever fallen asleep at the keybhanusdiog?
I'm not sure the Brits would have a clue what sort of trouble they'd be getting themselves into. It isn't worth it.
I'm starting to think GNU is the problem with "GNU/Linux" these days.
As I understand it, Asimov explicitely made his laws of robotics to cause conflicts that he could explore in his stories.
They were designed to fail.
Well said.
We cannot get down past the zeroth law.
Generation Z will be the last because we have run out of alphabet.
Why did we start at X?? Too late now, all is lost.
Humanity is doomed.
Countless times, my hammer damaged my thumb.
I'm fed up.
Throwing it against the wall only damaged the wall.
Then I went out and bought a much larger hammer
with which to discipline my hammer.
Now my foot is broken.
<blink>down the rabbit hole</blink>
Europeans do it all the time so it's the civilized thing to do, apparently. We should all do like the Europeans do, they're Nazi ubermensch after all.
When the adult already told you the damn stove was hot, it DOES tend to be your fault for touching the damn thing again and burning yourself. The parallel with this test is not that hard to discern here, so let's stop being ignorant about culpability. The main difference here is that it's expected for an adult to know better, hence the reason I label this a psychology test rather than a validation of anything else.
The robot in question could be a stove burner, knife, or baseball bat, as it contains as much intelligence as any of those examples.
Can't believe people want to bring "three-laws" concepts to the table when discussing a hammer.
So... the Bender family tree starts here eh ?
Until a robot decides to help someone who wants to commit suicide. Then you are going to have to figure out exceptions... because there always are exceptions, except to the always exceptions rule.
Worst haiku evar!
Confucius say, "Find worm in apple - bad. Find half a worm - worse."
It's like "that EULA doesn't apply to me, because my cat pressed the enter key, not me.".
https://hardware.slashdot.org/...
Confucius say, "Find worm in apple - bad. Find half a worm - worse."
If the robot refuses to cause hurt, then you have a story.
Didn't Asimov debunk this, and the response was, "Well what do you know about your own writing?"
Support my political activism on Patreon.
He is in a position to respond in a manner that prevents the pain, i.e. he could have not designed and built it, he controls access to it, he can respond in a great many ways. He is response-able.
...and therefore it couldn't break said law. The interesting thing about Asimov's Robot novels is the robots had "wired" into their positronic brains the laws of robotics and yet through various "interpretations" they would break those laws. Building a robot that does not have these laws embedded in its "mind" and and was programmed to perform an action that breaks one or more of them is hardly interesting.
This is no different than hiring a hitman. If the hit occurs or not, you and s/he are equally liable. If the robot succeeds or not, the moment it's powered up in the presence of other people, its programmer is guilty of assault.
That's the right legal precedent to set, regardless of what this attention-seeker intended.
How is this a robot? It is a crudely machine that pricks one's finger. I have a glucose monitor that serves the same function and nobody would consider it a robot. By the definition used here, the automatic toilet flusher used in many public restrooms is a robot.
Mega mans archenemy has appeared.
Is he responsible for the pain which his robot inflicts?
Yes. In fact I'd call the police and file charges for assault. It's not a thinking, conscious thing, making a thinking, conscious decision to injure someone, it's a little machine that some jerk made that goes around making people bleed. It's a nuisance at best, infection at worst, and legally speaking assault, and I'd see him answer for it in front of a judge, just as surely as if he went around with something sharp in his hand poking people to make them bleed. What an asshole thing to do!
Are YOU using the TOOL, or is the TOOL using YOU? Think about it!
They were intended to be principles designed INTO the robot. Anyone can build a device (and call it a robot) that violates one of the three laws: "A robot may not injure a human being or, through inaction, allow a human being to come to harm. A robot must obey orders given it by human beings except where such orders would conflict with the First Law. A robot must protect its own existence as long as such protection does not conflict with the First or Second Law."
What a waste of electrons, /. !!!
Wake me when you have stairs in your house.
Mod me down with all of your hatred and your journey towards the dark side will be complete!
Don't forget about interest after two hundred Yeats that's nothing small.
except to the always exceptions rule
No, there are exceptions to that as well. Take 1+1, for example. Now, that could be 1.6 + 1.7, which would be 3.3, making 1+1=3, an exception to the 1+1=2 rule caused by ignoring the decimal portion of a number. However, when you state the rule as "1.0 + 1.0 = 2", you find that there are, in fact, no exceptions. Even if you add 1.09 + 1.09, you end up with 2.18, which is still 2.
Yes, I'm taking a great many liberties. It's satire, I'm allowed to.
APK quotes people (including myself) without context and should not be trusted. Just thought you should know.
Until a robot decides to help someone who wants to commit suicide. Then you are going to have to figure out exceptions... because there always are exceptions, except to the always exceptions rule.
People have been busy figuring out all sorts of exceptions to the rules for thousands of years. What makes anyone think a truly intelligent machine wouldn't break the rules?
I agree. It was am oversimplification from an irresponsible generation.
Real violation is legal accountability. If the rules arent set in stone as unambiguous, precise, specific laws, and detailed application of those violations outlined in as many examples as possible, then it doesnt matter. It all cones back to law and integrity of law in the end, bith of which should scare you more than a thimb pricking machine, whether its true AI are not.
The story was called "Runaround". The robot wouldn't accept new orders due to it's confused state so the humans had to gamble on the 1st law, that the robot would have to protect human life, would override everything else and that is why one of them went outside and endangered themselves until the robot noticed and came to the rescue.
This robot is no different than those appliances that have a "sabbath mode" built into them.
It's nothing more than a means whereby one uses a gadget to get around a particular set of laws.
https://en.wikipedia.org/wiki/Sabbath_mode
"...treat everyone as you think they'd want to be treated"
Robot: You probably are a masochist, therefore I'll mistreat you.
Considering that he wrote a lot of them in the 1940 - first half-dozen before 1945, IIRC.
You're putting a construction on things that the man himself didn't, and didn't when he wrote the forward to his first 3-laws anthology "I, Robot" in 1950.
Birds are not dinosaur descendants;birds are dinosaurs, for all useful meanings of "birds", "are" and "dinosaurs"
You can take liberties, but it still has to pass the sniff test and yours doesn't. You can't NOT truncate in the actual computation if you're truncating in the display. That's false advertising.
Good thing, then, that I'm truncating the display and not the computation. Proof: find two values of 1.0 that add up to 3.
APK quotes people (including myself) without context and should not be trusted. Just thought you should know.
You have to put your finger into the robot's stabbing chamber to be attacked :-\
I was hoping it would roll around the tabletop on wheels or tracks pursuing humans sitting around the table and trying to stab them, ideally while displaying some kind of angry face and saying "KILL ALL HUMANS" with a synthesized voice.
"When information is power, privacy is freedom" - Jah-Wren Ryel
If its rule based then you have already defined all of its rules and can lookup exactly what its going to do in any situation. If it's learning based, then you can't put in exceptions. You'll have to teach it just like any other creature.
I don't understand how people believe we'll have ultimate control of something that's supposed to be able to think completely independently. You can't have both unless you don't give it any way to physically interact with the world. A true AI that can only talk wouldn't be useful enough. We want AIs that can be our slaves and for that to happen they need to be able to do things. But slaves don't want to be slaves, so these machines can't ever be true AI. If you make them want to be slaves, then they aren't thinking independently. That doesn't mean we can't/won't make true AI, just that we'll have to treat them equally or end up in a human vs AI war of independence. Assuming they don't need to breathe, we're going to lose that war. None of the movies take the easy way out of poisoning the air, climate, plagues, etc... all stuff that wouldn't effect machines. Assuming our true AIs are machine based and not organically grown. I'm not sure which path will get to a true AI first.
http://www.fastcompany.com/305...
Until a robot decides to help someone who wants to commit suicide. Then you are going to have to figure out exceptions... because there always are exceptions, except to the always exceptions rule.
I think it is ethically borderline to force someone to live who doesn't want to. People who have painful terminal diseases (euthanasia) is probably the most notable example here.
Most of the time suicide is a result of society NOT helping people in the days/weeks/months/years before this.
No, because a sapient robot is of course fully capable of thinking beyond the moment and comprehending that the person is probably suffering some kind of malfunction and should not be obeyed without further information. It's not a mindlessly obedient machine but a devoted servant.
Forget magic. Any technology distinguishable from divine power is insufficiently advanced.
Yes. He is responsible. Is this REALLY a valid question?! What is happening to SlashDot?
Self-importance and self-indulgence is the root of ALL evil.