Should Robots Feel? Jason Nemeth March 4, 2001 Philosophy 362

Size: px

Start display at page:

Download "Should Robots Feel? Jason Nemeth March 4, 2001 Philosophy 362"

George Anthony
6 years ago
Views:

1 Should Robots Feel? Jason Nemeth March 4, 2001 Philosophy 362 The purpose of this essay is to examine whether or not there would be practical reasons for creating a conscious, emotional machine. I will not delve to deeply into whether or not it is possible to create such a machine, as the argument as to what exactly would constitute a living conscious machine seems largely unsettled. Rather I will concentrate on whether or not we should create such a machine, if the possibility becomes available to us. Are there uses for such a machine that could not be satisfied by a complex automaton? Is there anything about real emotional response that would be necessary for a machine to operate autonomously, and still interact with human beings? What are the dangers? What are the ethical ramifications? It is questions such as these that will be the interest of this paper. Simulation and Replication An important idea central to this discussion is that of the dichotomy of simulation or replication of a human, by a robot. By simulation, I mean a robot that is functionally the same as a human in its behavior, but is still to be considered simply a machine. A robot that will react to inputs from its environment exactly the same as a person would, but is not itself a person, only a clever imitation of one. By replication, I mean a robot that actually is a person, not just an imitation of one. A strong AI intelligence that has a real, conscious mind, and emotions, just like a human being. Both of these arguments can be looked upon as controversial, in regards to the argument over whether or not a replication AI can actually be built. Few would argue that an excellent simulation of a human would not eventually be built. Sooner or later, as technology advances, it is highly likely a simulation that can pass for human will be devised. What seems to be controversial about this idea of simulation is that some in the field of AI feel that there would be no difference between a perfect simulation of a person and a replication of a person. That is, a functionally perfect simulation would be a replication. This seems to me somewhat of a behaviorist view, one in which only the inputs and outputs are important, not the mysterious inner workings of the person or the machine. Replication is perhaps more hotly contested, as many believe it is impossible for a computer, no matter how advanced, to ever truly replicate a person. Folk psychologists, as well as philosophers, believe that computers lack

2 the ability to reproduce some fundamental properties of the mind, such as semantic content, or intentionality. John Searle s Chinese room thought experiment (Minds, Brains and Science, p. 32) demonstrates the digital computer s lack of ability to have semantic content, something he considers quite central to the conscious mind of humans. To those on this side of the AI argument, no computer or robot can be a true replication, no matter how good a simulation it is. Although there continues to be much debate about these possibilities for AI, for the purposes of this paper I will assume, or imagine if you like, that both true replication and perfect simulation are, or will be at some time in the future, possible to achieve. I think that perhaps the more interesting question is which one of these possibilities is more desirable in the machines we will design and build. Emotional Response and Practicality Before considering whether or not robots should have emotions, I think it will be useful to take a cursory look at least into why human beings have emotions. I think there are a couple of ways of looking at this. First of all, human emotions are integral to our lives, the way we define ourselves as persons. Our personalities are closely connected and determined by the things we love and hate, our fears and our search for happiness. We value our emotions as intrinsic to our beings, even when they cause us pain, fear, or anxiety. That said, another way of considering emotions is to ask how they are functionally beneficial to us. It is not impossible to imagine human beings without emotions, cold calculating beings still fully capable of forming communities, procreating, and providing for their own survival. So is there a practical component to emotions, relating to our survival, our society? Some would say there must be, as we are evolved beings, and there must have been some advantage for us to have the emotions we do. This is of course fallacious, our emotions may have been as significant as our appendix to our survival, and even our selfawareness could conceivably be a byproduct of other developments, from an evolutionary standpoint. However, emotions can and do play a role in our decisions and actions, often ones that greatly affect our live and our death. The question is, do emotions serve a functional purpose that cannot be satisfied by a purely reason-based process? Philosophical psychologist Patricia S Greenspan describes emotions as propositional feelings, with important differences from more rational beliefs and judgments (Emotions & Reasons, p. 4):

3 (1) Although its appropriateness may be explained in terms of belief warrant, the evaluative component of emotions need not rest on reasons adequate for belief. (2) The affective component of emotions gives them a special role to play in rational motivation, as extrajudgmental reasons for action. Because of this, imagination and intuition can play a role in our actions, as these extrajudgmental emotions allow them as reasons for action. Greenspan also describes emotions as resistant to direct rational control, saying..it will assume resistance to direct control, of the sort we have over action. In this respect and others emotion seems to stand in between action and belief, exhibiting some features of both categories. (p. 10) From this I think we can determine some practical uses for human beings to have emotions. Emotions may help us to take action in situations where we may feel something that we can t justify through conscious rational beliefs. For example, in negotiating a business deal, one might feel an emotion of dislike or distrust for other parties involved, even when there is no clear rational reason to do so. This extrajudgmental feeling may be a distinct advantage, causing us to be wary of the other individual. Emotions that have their roots in imagination may play a practical role for us as well. The glossing-over of a loved ones imperfections is a good example of this. We can imagine an ideal version of a loved one, and this gives rise to emotions of love, caring, etc. that helps to hold relationships together, and certainly can play a role in propagation of our species. The Danger of Emotion After looking at some practical applications for emotion, some of the dangers inherent in emotion should become readily apparent. While extrajudgmental reasons for action may be useful in certain situations, they are certainly not infallible, and it may be extremely difficult to distinguish between intuitive or imaginative roots of an emotion. Emotions can and often do lead people to take unwarranted action. A perfect example of this is jealousy. A jealous husband, for instance, might go so far as to murder his wife if he suspected her of infidelity, even though his suspicions may not be rooted in rational beliefs, but rather in imaginative emotion. This is of course an extreme example, but it does effectively illustrate how dangerous emotion can be.

4 It is the dangers, the unpredictability of emotion, which leads many to fear the creation of emotional robots. Science fiction abounds with stories warning of the dangers of an emotional robot. For instance, HAL 9000, the conscious computer in Arthur C Clarke s book 2001: A Space Odyssey becomes paranoid, afraid for his life, leading him to murder the crew of the ship he controls. Another science fiction author, Isaac Aasimov, envisioned such robots being kept in check by his famous Three Laws of Robotics, central beliefs programmed into the robot to prevent it from harming humans. However, human emotions have the ability to circumvent even our most basic beliefs (consider the jealous husband), so one could easily argue that an emotional robot might ignore such laws when in a highly emotional state. It could clearly be argued that replications, truly emotional robots, should never be built because of this dangerous unpredictability. There are though, I think, some mitigating factors involved here, which I will discuss below. Societal Integration Human beings are of course subject to the same dangerous emotional unpredictability that replications might be. However, while there are many humans that take severe, unwarranted emotional action, most do not go so far as to hurt other human beings. We are all integrated into a society, which teaches us morals, ethics, and most importantly here, limits on the actions we take to express our emotions. The dangerous killer robots of science fiction tend to exist in a social vacuum, removed from human society and without a society of their own. A possible solution to the unpredictability of an emotional robot is social integration. Humanoid robots might be taught to feel as though they are part of the human race. Other emotional robots might be integrated into societies of their own, societies designed by man to have the same moral checks and balances ours do. A socially integrated replication would be no more likely to run amok then a socially integrated human. So the dangers of creating an emotional robot can be reduced so they are no greater than the dangers of creating a human being, possibly less, as robot societies could be carefully designed in a much more controlled way than ours are. However, we still have not looked at any practical reasons to build a replication. Humans may have practical uses for emotions, but is it practical for us to build robots with them? Robots in the Workplace From the standpoint of the businesses that employ them, it seems extremely unlikely we will find any

5 justification for emotional or conscious robots. It is often the very lack of human qualities that makes robots desirable in the workplace. Management professor Robert A Ullrich explains the advantages of working robots (The Robotics Primer, p42): Relaxations of environmental standards can provide substantial savings for the firm that substitutes robots for human labor. Robots can work around the clock and do not receive differential pay for night work; they are not prone to go on strike or to display other forms of labor unrest; they occasionally break down, but never get sick and are immune to the effects of drugs and alcohol; and they are readily available for work at unpleasant tasks for which human laborers are difficult to recruit. An emotional robot would be one with a mind like our own. With a sense of self-worth, a replication would likely desire the same benefits and payment that a human would. A robot with a desire for self-preservation and the ability to feel fear would not likely want to take on dangerous jobs any more than a human would. An imaginative robot would likely be prone to the same boredom at work that causes productivity loss in humans. At the same time, robots assigned specific tasks in the workplace should not run into situations in which extrajudgemental reasoning would be useful. While a sophisticated robot may need an extremely complex set of instructions to carry out assigned tasks, having clear logical rules for dealing with work related situations is a sufficient and productive way for such robots to operate. Human emotion and imagination may be important at certain levels of business, but this would not be necessary for robot workers, unless one wanted a business comprised entirely of robots, all the way up to the business owners! Social Interaction Perhaps the most likely place to find a practical use for an emotional robot is in the interaction between robots and humans. Already in today s world we can see the way people are embracing machines as something they can interact with on a social level. The popularity of electronic pets such as Gigapets, and kids attachment to these devices, is good example of this. Just last year, a popular Christmas present was a simple robot dog that would respond similar to a real dog to being pet, talked to, etc. It takes little imagination to see the logical progression of robotic pets to robotic human-like companions. It s clear that human beings would desire robots that are very humanlike in their actions and responses, especially in home applications like robotic nannies for children, or even automated customer service, such as a robotic phone answering system. But would these robots need to be truly conscious?

6 Would they need real emotions? In almost every circumstance, I think a human-like simulation would suffice for social interaction. Assuming a perfect simulation of a human could be built, this robot could be an excellent companion for a person while remaining a much safer robot to have around. Humans already demonstrate an ability to anthropomorphize all sorts of things; from the way they talk to their cats, to my blaming my car for its sadistic tendency to break down right after I blow my paycheck at the bar. Even with the knowledge that the robots we interacted with did not really feel, I think most people would still be happy with simulations as companions. Only building simulations would eliminate all the risks of an emotional robot deciding to harm a human, and we still haven t discussed any practical reason to have a true replication. I will discuss a reason now. There is one social interaction-type application for which I think a true replication would be highly desirable. This is the love-companion robot, surely a robot that might be desired when the technology becomes available. I am not speaking simply of a robot designed for sexual interaction with humans. A simulation would likely make a better simple sex robot. However, a robot meant to be a true life-partner for a human might need to go beyond simulation to truly satisfy that person s need for love and companionship. Even if a person had a perfect simulation of another human being, they would always know that the machine did not truly love or desire them, because the machine would not truly love or desire anything. Relationships based on simulated love and desire occur between humans already, and are generally ill fated. It becomes clear that a person might desire a robot with true feelings, emotions that, while artificially created, are not just simulation but real love. Whether or not these human/replication love relationships would work have any higher success rate than between two humans would be an entirely different matter, but we can at least now identify a practical reason to build an emotional robot. Slaves and Masters: Ethical Considerations Now that we have found a reason to build a replication of a human, I m going to go into reasons why we should not, from an ethical standpoint. Even if other important practical uses for replications can be found, I think the following line of reasoning will stand. Replications should not be built because of the important ethical distinction between a simulation and a living, feeling, sentient being. Computer Science professor Joseph Deken describes autonomy as the hallmark that would elevate robots to the status of a species (Silico Sapiens, p196) but this is not meant to imply that autonomous robots would be worthy of the same ethical considerations as sentient humans. A simulation, even a perfect one, or any other advanced AI would

7 still be a non-living machine, even with the ability to operate and reproduce independently of humans. Deken states that: (p.14) whatever forms our future robots take, we cannot let those forms induce us to accept the anthropomorphism that robots are our equals. Robots are not like humans nor worthy of the same considerations, no matter how sophisticated they become. They are our tools and creations, to be kept in place as a subservient species by whatever methods we find necessary. This sort of view is fine ethically, so long as machines never get beyond the level of a simulation. Machines do not need to be considered ethically, so we can treat them however we wish; use them and break them, make them do the jobs that we don t want to. However, as soon as a replication of a person is to be built, an important ethical boundary line is crossed. How is an emotional robot to be treated? Once it has been built, is it to be bought and sold, commanded to work by its owners? If a robot were created that is truly a person, a relationship such as this would be one of slave and master. We all already know that it is highly unethical to treat a person in this way, but it was only a short time ago that western peoples stopped using humans as slaves, and other societies around the world still treat some groups of people in this way. Consider Deken s words above again. It s likely very similar words were used by Americans to describe their African slaves just a couple hundred years ago. Even if this ethical consideration was well understood by the designers and builders of an emotional robot, there is no guarantee that others will view the machine as a person. Many people would stick to their belief that a real emotional robot is impossible, and Deken s words, to be kept in place as a subservient species by whatever methods we find necessary might well be the cry of non-believers. A new era of slavery and degradation for a sentient species could be beginning. To go even further, even the acceptance of replications as real persons does not mean we would consider them equals. Such robots would likely finds themselves taking the place women are often in, considered very much human, but subservient to others. The love-companion robot I discussed earlier might truly love its human companion, only to find itself a subservient whipping boy for that human. Perhaps these problems will not stand forever. It is true that humanity has made important leaps forward in ethical treatment of other human beings over the last century, improving racial and gender equality to some extent. Still, we have along way to go in this endeavor, and it seems impossible to imagine treating other species ethically

8 until we can do it with our own. Luckily, we can simply avoid the ethical problems of how to treat emotional robots until we have solved the problems of slavery and intolerance, and other unethical ways of treating people. We can simply not build any. Works Cited 1. Searle, John, Minds, Brains and Science. Cambridge,MA: Harvard University Press, Greenspan, Patricia S., Emotions and Reasons: An Inquiry into Emotional Justification. London, UK: Routledge, Ullrich, Robert A., The Robotics Primer: The What Why and How of Robots in the Workplace. New Jersey: Prentice-Hall Inc., Deken, Joseph, Silico Sapiens: The Fundamentals and Future of Robots. Toronto: Bantam, Liukkonen, Petri, Books and Writers/Author s Calendar/Isaac Asimov (3 March 2001) 6. Clarke, Arthur C., 2001:A Space Odyssey. New York: Harper, 1968

Should AI be Granted Rights?

Lv 1 Donald Lv 05/25/2018 Should AI be Granted Rights? Ask anyone who is conscious and self-aware if they are conscious, they will say yes. Ask any self-aware, conscious human what consciousness is, they