Nao, manufactured by Aldebaran Robotics, is the first robot to have been programmed with an ethical principle.
Image: Darren Braun
In Brief
- Robots that make autonomous decisions, such as those being designed to assist the elderly, may face ethical dilemmas even in seemingly everyday situations.
- One way to ensure ethical behavior in robots that interact with humans is to program general ethical principles into them and let them use those principles to make decisions on a case-by-case basis.
- Artificial-intelligence techniques can produce the principles themselves by abstracting them from specific cases of ethically acceptable behavior using logic.
- The authors have followed this approach and for the first time programmed a robot to act based on an ethical principle.
In the classic nightmare scenario of dystopian science fiction, machines become smart enough to challenge humans—and they have no moral qualms about harming, or even destroying, us. Today’s robots, of course, are usually developed to help people. But it turns out that they face a host of ethical quandaries that push the boundaries of artificial intelligence, or AI, even in quite ordinary situations.
Imagine being a resident in an assisted-living facility—a setting where robots will probably become commonplace soon. It is almost 11 o’clock one morning, and you ask the robot assistant in the dayroom for the remote so you can turn on the TV and watch The View. But another resident also wants the remote because she wants to watch The Price Is Right. The robot decides to hand the remote to her. At first, you are upset. But the decision, the robot explains, was fair because you got to watch your favorite morning show the day before. This anecdote is an example of an ordinary act of ethical decision making, but for a machine, it is a surprisingly tough feat to pull off.
This article was originally published with the title Robot Be Good.
Already a Digital subscriber? Sign-in Now
If your institution has site license access, enter here.
See what we're tweeting about
15 Comments
Add CommentThis has already been done...a few times.
Reply | Report Abuse | Link to thisJohn E. LaMuth, has a US Patents for it, the website is here;
http://www.angelfire.com/rnb/fairhaven/patent.html
http://www.google.com/patents?vid=USPAT6587846&id=iRUPAAAAEBAJ&dq=6587846
I've heard of others...I would like to know how they all 'stack up' so to speak...
I can't wait!
Writing a moral principle and acting morally are two different things. Acting morally requires a free will, which is lacking in a robot that acts in a binary logic.
Reply | Report Abuse | Link to thisA robot that calculates the probabilities and act according to the result that corresponds to the moral principle relevant to that case is still a machine that acts "like" a moral agent, but not a moral being that acts morally. As humans know we act morally not because calculations show that way but because we "feel" that it is the right thing to do in that instance.
Reply | Report Abuse | Link to thisIn “Robot Be Good” authors Anderson note that some ethicists find a flaw in the Mill/Bentham doctrine of Hedonistic Act Utilitarianism, a theory which holds that the ethicality of an act can be objectively calculated according to a weighing of the pleasure/displeasure ratio of those affected. Objectors note that such a calculation might lead to an individual being sacrificed for the good of the majority, hardly an ethically acceptable outcome when large numbers receive small benefit while one individual suffers greatly.
Reply | Report Abuse | Link to thisHowever, the authors go on to say that the Mill/Bentham doctrine “at least demonstrates that a plausible ethical theory is, in principle, computable.” I’m not at all sure how one finds plausibility in a methodology that has such an obvious flaw. The fact that I might theorize that I can fly by flapping my arms does not elevate my theory to plausibility even “in principle.” Typically, in fact, the presence of such a fundamental deficiency renders a theory IMplausible.
That a flawed theory exists demonstrates only that the human mind can invent flawed theories. To claim for it any more validity than that is to flirt with something like a tautology.
I imagine that the authors cited the Mill/Bentham methodology as a way of suggesting that a formulaic approach to calculating ethicality is an avenue worth exploring. In no way, however, does it “demonstrate” plausibility or possibility, even “in principle.” The reference to Mill/Bentham would be more appropriately employed in an argument against the practicality of ethical formulas.
My ethical issue isn't for the 'bot...it seems questionable to me to have a feature article about a product advertised on the page page of the magazine.
Reply | Report Abuse | Link to thisEveryone has different ethics - thehistoryof mankind from Biblical day on
Reply | Report Abuse | Link to thisSex
Abortion
Stem cells
War for what?
Terrorists
even Hitler
Do we program to only that which WE believe is ethical?
Where fore the concept of ethics and morality?
The issue of ethics and morality has been considered in a formal and computational manner by the Artificial Intelligence community for a few years now. Some of the most recent works that show how to specify and implement the moral reasoning mechanisms can be found in scientific publications such as
Reply | Report Abuse | Link to thisL. M. Pereira, A. Saptawijaya, "Modelling Morality with Prospective Logic", in: International Journal of Reasoning-based Intelligent Systems (IJRIS), 1(3/4):209-221, 2009.
And I'm sure there might be many other researchers currently investigating in this area.
Free will is a construct that exists to justify a system of evaluating beings on their moral achievement.
Reply | Report Abuse | Link to thisEthical robots, making decisions based on pre-defined ethical variables are not moral agents. They are calculators. The morality of their actions exists only in the minds of their programmers and users. The programmer's goal is to satisfy the needs of the situation, i.e., leave the users feeling that the robot acted ethically, NOT TO GET THE ROBOT INTO HEAVEN! LOL.
Surely managed care support functions, for example, might necessitate autonomous control of robots' site navigation without enabling some however limited set of algorithms to make potential quality of life or death determinations concerning patients.
Reply | Report Abuse | Link to thisIf this society can only relinquish moral responsibility for personal care decisions to the product of a machine programmer then I prefer to chose my own fate, thank you. Institutional health care is already impersonal and desensitized enough without implementing robotic personal care.
"Writing a moral principle and acting morally are two different things. Acting morally requires a free will, which is lacking in a robot that acts in a binary logic."
Reply | Report Abuse | Link to thisDo we really have any more "Free will"? We are just acting in biochemical and electrical based logic. I also think that acting in concordance with a moral rule (acting morally) doesn't require anymore "Free will" then present AI possesses.
"A robot that calculates the probabilities and act according to the result that corresponds to the moral principle relevant to that case is still a machine that acts "like" a moral agent, but not a moral being that acts morally."
So humans don't weigh the moral consequences of their actions in relation to principles and rules they were taught as children and then come to a conclusion upon which they act? Evidently I missed that memo.
"we act morally not because calculations show that way but because we "feel" that it is the right thing to do in that instance."
I'd question whether or not morality based on nothing but feelings, as you seem to think is the way humans make moral decisions, has any true value. I know in my own case I attempt to weigh the consequences of actions based on their positive versus negative results for a given system. Beyond that I would suggest that our feelings are a primitive form of calculating moral correctness of an action generated via billions of years of evolution.
@gesimsek, "Acting morally requires a free will" well then humans don't act morally either. The evidence from cognitive science indicates that humans don't have free will. We are controlled by our genes, by our development and by our experiences. So at least we'll be on an even playing field.
Reply | Report Abuse | Link to thisCharacteristically, the consequences of human decisions and action include experiencing, long term, emotions such as regret and remorse, etc. Not suffering similar consequences qualifies programmable decision processors as sociopaths or psychopaths. Perhaps this is your personal frame of reference? Perhaps modern society needs a big hug!
Reply | Report Abuse | Link to thisI'm not sure I see where ethics come into the picture here. The kind of theoretical situation outlined in this article describes an extremely simple machine which hardly requires a complex "ethical" algorithm to derive its behaviours. We might as well design ethical rules for the computer guidance systems driving our 747s. Rather, this characterization of robot "ethics" seems to evoke visions of highly complex artificial intelligence--visions which are in noticeable conflict with the rather simple abilities and menial duties of the authors' hypothetical robot. I am sure most Asimov fans will recall that the three laws were designed for extremely human-like machines which imitated their creators to perfection. Of course, in science fiction the fulfillment of this anthropomorphic machine fantasy is perfectly reasonable. Meanwhile, in reality, it is rather unlikely that we will see such accurate mimicry within out lifetimes, if ever. As a result, I am only left to wonder why the authors of this article attempt to suggest that we need any kind of ethical code of conduct for our machines. Perhaps they too are deluded by childhood fantasies of metal men.
Reply | Report Abuse | Link to thisReally a educative and informative post, the post is good in all regards,I am glad to read this post.
Reply | Report Abuse | Link to thishttp://www.cellhub.com/t-mobile-cell-phone-plans/even-more-plans.html
Even More Cell Phone Plans
A video concerning the research can be seen here:
Reply | Report Abuse | Link to thishttp://www.youtube.com/watch?v=ZLdvCDFriTQ
A pdf concerning the science/philosophy behind the research is freely available here:
www.aaai.org/ojs/index.php/aimagazine/article/view/2065/2052