Correcting Asimov's 3 Laws of Robotics

  • Thread starter Thread starter Angainor
  • Start date Start date
Status
Not open for further replies.
A

Angainor

Guest
Isaac Asimov invented the 3 Laws of Robotics for his science fiction. The laws were intended that people were safe from harm from robots. I think he got the laws wrong. Here they are as he invented them:
  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 the orders given to it by a human being, except where that would conflict with the First Law.
  3. A robot must protect its own existence, except where that would conflict with the First or Second Law.
These laws caused lots of conflict in Asimov’s fiction as well as the recent movie I, Robot (with Will Smith). Most of the conflict comes when the robots try to save people from themselves and turn into tyrants. Here is how I would fix the laws:
  1. A robot may not injure a human being.
  2. A robot must obey the orders given to it by a human being, except where that would conflict with the First Law.
  3. A robot must protect its own existence, except where that would conflict with the First or Second Law.
This would end meddling by the robots and robots would be just tools as they were meant to be. You could still give robots specific standing orders, such as, “if a grand piano is about to fall on my head, save me”.

If you are not entirely comfortable with that, I would consider the following alternative:
  1. A robot may not injure a human being.
  2. A robot must obey the orders given to it by a human being, except where that would conflict with the First Law.
  3. A robot cannot, through inaction, allow a human being to come to harm, except where that would conflict with the First or Second Law.
  4. A robot must protect its own existence, except where that would conflict with the First, Second, or Third Law.
This would allow some meddling, unless ordered not to by a human being. This would have solved Will Smith’s problem in the movie where the robot saved him in a car crash instead of a little girl even when Will ordered it to save the girl.

So, what does everybody think?

P.S. Asimov himself tried to fix the laws by adding a Zeroth law. I think this fix represents the darkest of modernist evils since it puts the interests of “humanity” over the welfare of individuals.
  1. A robot may not injure humanity or, through inaction, allow humanity to come to harm.
  2. A robot may not injure a human being, or through inaction, allow a human being to come to harm, except where that would conflict with the Zeroth Law.
  3. A robot must obey the orders given to it by a human being, except where that would conflict with the Zeroth or First Law.
  4. A robot must protect its own existence, except where that would conflict with the Zeroth, First, or Second Law.
 
Angainor,

Your post reminds me of a technical paper I read once on placing GPS-style beacons in the asteroid belt to allow spacecraft flying around the solar system to know where they are at all times. It may someday be an issue, but it certainly isn’t one yet.

I agree with you that Asimov’s Zeroth Law is indeed the darkest of modernist evils. “Humanity” is an abstract concept; people are real. Perhaps a better alternative is “Given a choice between an action that will harm more people and an action that will harm fewer people, the robot will not choose the action that harms more people.” But even that is not perfect; there are degrees of harm, for example, and allowing one person to die so that a hundred people will not lose fifty dollars each is not right.

I’ve not seen the Will Smith movie, but from your description it appears that the robot’s actions in the car crash did not even follow the Three Laws as stated. The little girl in question was a human being, after all, and not saving her was a violation of the First Law.

While your amended laws are an improvement, I don’t think they are a complete fix. They would result in robots spending all their spare time doing things to save human beings from harm, and I think you would be right back to the tyrannical nursemaid situation.
  • Liberian
 
40.png
Liberian:
Angainor,

Your post reminds me of a technical paper I read once on placing GPS-style beacons in the asteroid belt to allow spacecraft flying around the solar system to know where they are at all times. It may someday be an issue, but it certainly isn’t one yet.

I agree with you that Asimov’s Zeroth Law is indeed the darkest of modernist evils. “Humanity” is an abstract concept; people are real. Perhaps a better alternative is “Given a choice between an action that will harm more people and an action that will harm fewer people, the robot will not choose the action that harms more people.” But even that is not perfect; there are degrees of harm, for example, and allowing one person to die so that a hundred people will not lose fifty dollars each is not right.

I’ve not seen the Will Smith movie, but from your description it appears that the robot’s actions in the car crash did not even follow the Three Laws as stated. The little girl in question was a human being, after all, and not saving her was a violation of the First Law.

While your amended laws are an improvement, I don’t think they are a complete fix. They would result in robots spending all their spare time doing things to save human beings from harm, and I think you would be right back to the tyrannical nursemaid situation.
  • Liberian
In the movie both cars went into the river and started to sink. Will Smith was in one and the little girl was in the other. When the robot arrived on the scene it calculated Will Smith’s character had a 41% survival chance but the girl only 11% so it chose to save the one with the higher chance of survival.
 
40.png
thistle:
In the movie both cars went into the river and started to sink. Will Smith was in one and the little girl was in the other. When the robot arrived on the scene it calculated Will Smith’s character had a 41% survival chance but the girl only 11% so it chose to save the one with the higher chance of survival.
Let’s say Will’s character had a 98% chance of survival, and the girl’s, 50% chance. Assuming these chances of survival represent the chances of survival given the inability of the robot to save either one, why wouldn’t the robot try to save the girl, since Will would have had a greater chance of survival without the aid of the robot?
 
40.png
Ahimsa:
Let’s say Will’s character had a 98% chance of survival, and the girl’s, 50% chance. Assuming these chances of survival represent the chances of survival given the inability of the robot to save either one, why wouldn’t the robot try to save the girl, since Will would have had a greater chance of survival without the aid of the robot?
Both of them in different cars were trapped inside. In both cases they would have died if the robot had not come along so their survival percentage depended on a rescue by the robot.
 
40.png
Liberian:
While your amended laws are an improvement, I don’t think they are a complete fix. They would result in robots spending all their spare time doing things to save human beings from harm, and I think you would be right back to the tyrannical nursemaid situation.
What about my first suggestion? I think with this set of laws, the robot would stand by while a human got hit by a bus.
  1. A robot may not injure a human being.
  2. A robot must obey the orders given to it by a human being, except where that would conflict with the First Law.
  3. A robot must protect its own existence, except where that would conflict with the First or Second Law.
 
40.png
Angainor:
What about my first suggestion? I think with this set of laws, the robot would stand by while a human got hit by a bus.
  1. A robot may not injure a human being.
  2. A robot must obey the orders given to it by a human being, except where that would conflict with the First Law.
  3. A robot must protect its own existence, except where that would conflict with the First or Second Law.
I think we still have a problem.

Take for example, the eugenics programs in the USA at the beginning of the 20th century and of the Nazis. Even slavery to an extent.

To do such things the defintion of what a “human being” is changed.
 
40.png
ByzCath:
I think we still have a problem.

Take for example, the eugenics programs in the USA at the beginning of the 20th century and of the Nazis. Even slavery to an extent.

To do such things the defintion of what a “human being” is changed.
Yes, the problem is a very basic one: the moment you have a sentient being deciding what is best for you, you become subject to and inferior to that sentient being. If that being is God, that’s fine; we are subject to and inferior to God. But if we were to put ourselves into the hands of beings that we have created, then we start to lose our humanity.

(off-topic)
This is probably what makes some men go camping now and then. They want to declare their independence from the everyday support systems that usually sustain them.
(/off-topic)
  • Liberian
 
Angainor,
I love books and movies about robots, Caves of Steel, etc, but hope that we never manage to create man-like ones. I suppose it’s only a matter of time, though, since they’ve already come up with serving and vacuuming robots.
We have enough problems trying to solve the problems caused by human beings’ greed and violence. Adding sentient machines to the mix will only compound confusion.
Imagine wanting to drive to church on an icy day, and your robot taking the keys away to protect you from yourself.
 
The problem with the robot is that it did too good a job of treating human beings as equally valuable.

It sought to make its best effort to save human life. It chose the one most likely to succeed.

The issue here is that Will Smith character valued the girl’s life above his own. The robot was not programmed to take that into account.

The robot was doing its job at improving the chance it would save a life. What if the robot had calculated that the girl had identically zero chance of being saved but Will Smith didn’t realize that? Would it still have been the “wrong” decision?

Actually what we have in this case is a totally and objectively pro-life robot, and does not discriminate. Smith preferred to sacrifice himself in order to give the girl a chance to live. Keep in mind that by the time Smith knew the reasons the robot saved him, the emotional impact of the moment had already passed. The robot may even have realized that Smith, when giving the order, did not have all the information available to make the decision as it did.

The “correct” solution to this particular dilemma would be for us to program a system into the robots whereas the robot could learn how to place a numerical value on different human beings so when a decision has to be made it can be weighted in favor of the preferred ones.

Perhaps the robot did the best thing in the long run. If the robot had saved the girl in the first place, then Smith would not have been around to save the rest of humanity. The girl was sacrificed for many, though not by her own will but circumstance.

We humans, especially macho men, like to say, “women and children first” and the traditional context is about who gets the lifeboats when there aren’t enough or when there isn’t enough time to all escape from whatever the problem isl. That’s good strategy for humanity, because it keeps the future intact. However, it is not an absolute law, in that if the men are the ones who know and have ability to lead, navigate, or whatever, then somebody should go help the women and children (if you want to view is less sexist then just say “children”). For example, we wouldn’t put a boat with 20 children under age 10 with no marine experience in favor of a boat with maybe 18 children and one adult to guide them?

On an airplane, one is admonished that if pressurization is lost, one must put on one’s own oxygen mask first so one doesn’t pass out while assisting others. This is the other side of “women and children first.” Take that as an absolute value system, and those who need assistance are not likely to be helped by serving first, as those assisting them will not be able to breathe.

The movie, I think, juxtaposes absolute and relative, in terms of behavior systems. If the robot behaves exactly as programmed, we don’t get what we want. We want to have human interpretation above and beyond any written rules so that in a critical situation we may override our previous notions of what good behavior is.

Personally I think this was the message of Christ. The law, as interpreted by Christ, was not hostile to humans but only to sin. The law, as interpreted by everyone else, was robotically applied (at least in some cases) and that’s what I think Christ was against. For example, if the law says one caught in adultery must die by stoning, then a robot would be obedient to the law, and that is the role the people took in the parable. Christ introduced human reason and compassion into the equation, which transcended written law.

To me, the message is that no law, no matter how simple and harmless it seems, if carried out with perfect obedience to the law as written, will be able to replace human love and compassion as decision making tools. There had to be a disobedient robot to break the chain of power, once the decision making process had led them into a position of control.

Strictly speaking, in human decision making, one may actually have to choose which of two pro-life actions to take. We want to have the flexibility to make a decision based on human intuition (or guidance of the Holy Spirit) that goes against all previous rules.

If Christ were a robot like the others, he would not have let his people pick and eat grain on Sunday. He would not have prevented them from stoning the prostitute. I think that is why we need to be careful not to live up to anti-Catholic’s claims of being “legalistic.” We must have an out for human love to make a decision based on the future situation, whether or not the scenario was envisioned in designing the laws with which we claim to live by. Remember that of faith, hope, and love, the greatest of these is love. The robot followed its rules in perfect “faithful” obedience, but did not try to factor in “love.” The rules it followed were human rules, designed by humans who didn’t realize what their own rules implied in a particular situation.

Alan
 
40.png
Viki59:
Angainor,
I love books and movies about robots, Caves of Steel, etc, but hope that we never manage to create man-like ones. I suppose it’s only a matter of time, though, since they’ve already come up with serving and vacuuming robots.
We have enough problems trying to solve the problems caused by human beings’ greed and violence. Adding sentient machines to the mix will only compound confusion.
Imagine wanting to drive to church on an icy day, and your robot taking the keys away to protect you from yourself.
That’s exactly what I mean. We do not want the program to follow rules without or ability to “opt out” of the rules based on human intuition or whatever causes one to go against prior instructions.

Basically, we don’t want robots – or people, for that matter – to blindly obey rules. Christ showed that rules are here to serve people and not the other way around.

Alan
 
40.png
Viki59:
Imagine wanting to drive to church on an icy day, and your robot taking the keys away to protect you from yourself.
Robot or human, that would upset me.
40.png
Viki59:
We have enough problems trying to solve the problems caused by human beings’ greed and violence. Adding sentient machines to the mix will only compound confusion.
Who is “we” and how and why are you trying to “solve” the problems? (Would you take my keys on an icy day?)

Depending on how people try to “solve” the problems they might be just as meddling as any metal men.
 
Viki59,

I just noticed you screen name. 🙂

The computer that managed the robots in the Will Smith movie was named V.I.K.I.

Just an observation.
 
40.png
AlanFromWichita:
The “correct” solution to this particular dilemma would be for us to program a system into the robots whereas the robot could learn how to place a numerical value on different human beings so when a decision has to be made it can be weighted in favor of the preferred ones.
Alan,
How could this be done, after all, isn’t all life valuable and equal?

No one life is more valuable than any other.

When we start to value one life over that of another is when we start down that slippery slope to dehumanize others. Isn’t this what eugenics and the Nazis movement is all about?
 
Ahh this will all be blow out the water once the military takes up the technology.
 
40.png
ByzCath:
Alan,
How could this be done, after all, isn’t all life valuable and equal?

No one life is more valuable than any other.

When we start to value one life over that of another is when we start down that slippery slope to dehumanize others. Isn’t this what eugenics and the Nazis movement is all about?
In the movie, Will Smith orders a robot to save a little girl instead of himself. The robot overrode Smith’s orders because of its interpretation of the robot rules, because it calculated the chance of saving Smith was four times the chance of saving the girl.

Thus Smith was bitter toward robots after that because he felt it was more “human” to save the little girl, even though her chances were slim.

My point is that in order to program a robot to make decisions the way Smith wanted them to be made, we would have to give them a weighting factor. Yes, it would be very scary for this to take place, but the premise of the movie is actually based on the robots taking over humanity in a violent confrontation.

Alan
 
40.png
Liberian:
Angainor,

I agree with you that Asimov’s Zeroth Law is indeed the darkest of modernist evils. “Humanity” is an abstract concept; people are real. Perhaps a better alternative is “Given a choice between an action that will harm more people and an action that will harm fewer people, the robot will not choose the action that harms more people.” But even that is not perfect; there are degrees of harm, for example, and allowing one person to die so that a hundred people will not lose fifty dollars each is not right.
  • Liberian
I forget the title, but Asimov actually confronted the “defining humanity” problem in a story toward the end of his career.
In the story two very advanced robots are conversing and they decide that rationality is the hallmark of humanity and since, of course, more rational thatn flesh & blood humans, they would choose to preserve themselves rather than their masters.

Asimov also touched on the different kinds of harm. In the original “I, Robot” anthology one of the stories was about a robot who was aware when people were suffering emotional harm and so had to lie to them.
 
The movie was completely innacurate to the book from my understanding, and therefore not written to coincide exactly with what Asimov might have chosen.
 
40.png
Arwen037:
The movie was completely innacurate to the book from my understanding, and therefore not written to coincide exactly with what Asimov might have chosen.
Yes, the movie was nothing at all like the book, which after all, started out as merely a collection of short stories.

Asimov’s 3 Laws is an interesting concept, but it well never be used in practice. The Roomba vacuum robot doesn’t have the 3 laws built in, and I suspect there are no engineers even thinking about this. My computer is sure lacking in the 3 laws; it’s internal laws are often at odds with its owner.
 
40.png
JimG:
Asimov’s 3 Laws is an interesting concept, but it well never be used in practice. The Roomba vacuum robot doesn’t have the 3 laws built in, and I suspect there are no engineers even thinking about this. My computer is sure lacking in the 3 laws; it’s internal laws are often at odds with its owner.
Yes, and think about it. In the books and the movie the
Asimov Laws were mostly a marketing ploy.

Buy our robots for your home, your business, your industry. Don’t fear them, use them - they are “3 Laws Safe”

The Military or private industry doesn’t need to apply any marketing ploys if they create robots.

theinquirer.net/?article=30306
 
Status
Not open for further replies.
Back
Top