Love and Sex with Robots_The Evolution of Human-Robot Relationships

by David Levy

  The initial forays by roboticists into the world of fully interactive autonomous robots focused on entertainment, with creations such as robot toys, robot pets, and robots that play sports. Simple electronic cats and dogs have been shown to provide psychological enrichment for humans, being both pleasurable and relaxing to play with. More recent research has started a trend for interactive robots that act as human helpers, showing visitors around museums, caring for hospital patients and the elderly, and providing therapy to cope with emotional problems. Japanese researchers have shown, for example, that the mood of a child can be improved by interaction with a robot and that robots are able to encourage problem children to communicate more with each other and with their caregivers.

  Toys such as Furby, Tamagotchi, and Robosapien and the virtual characters that inhabit the worlds created by computer-game designers are part of an evolutionary technological process that has turned the simulation of cognizance and perception into something much more—a force with massive potential. This force already manifests itself as an expectation, by many people, that the toys and computer programs with which they interact today will exhibit signs of life. We know that it is an artificial form of life, but the expectation that something will exhibit even an artificial form of life is a significant step toward the acceptance of such forms as real. And that day is not as far into the future as you might believe. To some extent at least, the acceptance of robots as entities capable of interesting, useful, and rewarding interaction with humans has already arrived. In Chevy Chase, Maryland, a nonprofit organization called the Institute of Robotic Psychology and Robotherapy has been set up to study some of the fundamental questions of mind, emotion, and behavior that relate to human-robot interaction. Robotic psychology focuses on human-robot compatibility, while robotherapy concentrates on the task of employing interactive robots as therapeutic companions for people who have psychological problems or are handicapped physically, emotionally, or cognitively.

  The current state of the art in robotics and in other domains within artificial intelligence is not what this book is about; it is merely the starting point for my thesis. We already have android robots,* whose appearance is designed to resemble humans, such as Honda’s ASIMO, Waseda University’s WABOT, and Toyota’s trumpet-playing robot.† Other robots that have already been built include Volkswagen’s Klaus, which can drive a car; robots that can mow our lawns and vacuum our carpets; robots that can change their own shape in order to maneuver through disaster sites in their search for victims; and robots that can reproduce—picking up and assembling the pieces of exact replicas of themselves. And already we have computer software that excels in many intellectually demanding tasks and in most areas of creativity, and we have software that can exhibit humanlike emotions.

  How all these feats, and many others in AI, have been accomplished, is explained in my earlier book Robots Unlimited, where I also summarize the technologies that will make possible remarkable advances in the power and speed of computer processing during the decades to come, technologies such as DNA computing, quantum computing, and optical computing. When these new computer technologies have been developed to maturity, and when they have been combined with what will then be the latest advances in AI research, the intellectual capabilities and the emotional capacities of robots will be nothing short of astounding. They will look like humans (or however we want them to look). They will be more creative than the most creative of humans. They will be able to conduct conversations with us on any subject, at any desired level of intellect and knowledge, in any language, and with any desired voice—male, female, young, old, dull, sexy. The robots of the mid-twenty-first century will also possess humanlike or superhuman-like consciousness and emotions.

  As the potential usefulness of robots began to be debated, alongside a discussion of the many tedious tasks that humans would delegate to machines rather than perform themselves, it was realized that the diversity of human activities needs a diversity of assistants, with different robots designed to perform and solve different tasks. Robots would be needed in industry to operate machines; they would be needed by the military and the rescue services to help at disaster sites; they would fulfill a role as replacement or adjunct teachers; they would diagnose illness and assist in the operating room. These and many other tasks soon became areas of research for roboticists.

  The tasks that the early robots were designed to help solve had little to do with human emotions, and therefore they did not require any emotional response from a robot. But as psychology and cognitive science began to be studied in relation to robots, it became apparent that we need to consider what relationships might one day develop between man and machine, between human and robot. Suddenly it was important to think about what might happen when a robot communicates with a human on a personal level rather than merely for pragmatic reasons linked to the robot’s “mechanical” functionality. It was no longer enough for the human to press a button or say, “Please bring me a cup of tea,” and for the robot to do as requested. Instead a new generation of AI researchers was investigating more meaningful relationships between humans and what Alexander Libin has called “artificial partners.”

  Again it is the Japanese robot scientists who have led the research in “partner robots,” recognizing that “robots increasingly have the potential to interact with people in daily life. It is believed that, based on this ability, they will play an essential role in human society in the not-so-distant future.”4

  There are those who doubt that we can reasonably ascribe feelings to robots, but if a robot behaves as though it has feelings, can we reasonably argue that it does not? If a robot’s artificial emotions prompt it to say things such as “I love you,” surely we should be willing to accept these statements at face value, provided that the robot’s other behavior patterns back them up. When a robot says that it feels hot and we know that the room temperature is significantly higher than normal, we will accept that the robot feels hot. When it says that the piano is being played too loudly, recognizing of course that it is listening to a piano, we will accept that the music is too loud for the robot if it also sounds loud to us. Just as a robot will learn or be programmed to recognize certain states—hot/cold, loud/quiet, soft/hard—and to express feelings about them, feelings that we accept as true because we feel the same in the same circumstances, why, if a robot that we know to be emotionally intelligent, says, “I love you” or “I want to make love to you,” should we doubt it? If we accept that a robot can think, then there is no good reason we should not also accept that it could have feelings of love and feelings of lust. Even though we know that a robot has been designed to express whatever feelings or statements of love we witness from it, that is surely no justification for denying that those feelings exist, no matter what the robot is made of or what we might know about how it was designed and built.

  The mere concept of an artificial partner, husband, wife, friend, or lover is one that for most people at the start of the twenty-first century challenges their notion of relationships. Previously, the relationship between robot and human has always been considered in terms of master to slave, of human to machine. But with the addition of artificial intelligence to the machine-slaves conceived in the twentieth century, we have now made them into something much more. Yes, they might still be programmed to do our bidding, yet they are also being programmed to consider not only our practical wishes, serving drinks and mowing the lawn, but our feelings as well. By endowing robots with the capability of communicating with us at a level we can understand, a human level, and by building robots that have at least some appearance of humanlike features, we are rapidly moving toward an era when robots interact with us not only in a functional sense but also in a personal sense.

  In the middle of the twentieth century, the founding father of the science of cybernetics,* Norbert Wiener, extolled the virtues of the interactive robots of the future, including their ability to learn from experience and, as a result of what t
hey learn, to improve the lot of those with whom they might be interacting. The psychological benefits of such robots would, he asserted, be similar to the psychological benefits that the cared-for receive from their human carers. But to gain acceptance by the humans with whom they are interacting and for whom they are caring, robots need to imitate at least some of our social cues and to be at least vaguely similar to us in appearance.

  Take a look at the Kismet robot designed and built at MIT by a team led by Cynthia Breazeal. It has a head, as we do; it has eyes, as we do; it has a mouth with moving lips, as we do. Put simply, a human interacting with a robot will be more at ease if the robot exhibits some human appearance and characteristics than if the robot is merely a metal box with wires, lights, and wheels. The more humanlike a robot is in its behavior, in its appearance, and in the manner with which it interacts with us, the more ready we will be to accept it as an entity with which we are willing or even happy to engage.

  KISMET

  For this reason certain trends in toy design can be viewed as precursors to twenty-first-century android robot designs. While remarkable advances were seen in robotics research during the latter decades of the twentieth century, the cosmetic appearances and forms of dolls and similar toys have been part of a less dramatic but nevertheless important trend in design. Even in that iconic product the Barbie doll, one can see breasts, while some other dolls, intended for older children and young teenagers, are marketed with a line of seductive-looking lingerie. There have also been boy-shaped characters with prominent penises, marketed as props for use in sex education.

  The benefits for human-robot interaction of the human’s familiarity with the robot’s appearance and behavior are mirrored in the relationships between many humans and their pets. The human-pet relationship is also a kind of partnership, with some parallels to certain aspects of human-human relationships. It is a partnership that was enthusiastically seized upon by robot designers in the early days of recreational robots. In the case of traditional family pets—cats, dogs, rabbits, and the like—our relationship partnerships with those animals create a measure of emotional attachment and have been shown to be of therapeutic benefit to us. To make the partnership work with robots, designers have created robot dogs such as Sony’s AIBO, robot cats, and other animal-like robots such as Furby, which sold more than 40 million pieces. Furby starts out life talking in a gibberish language called Furbish but with time reduces the incidence of Furbish in its vocabulary and correspondingly increases its use of English or whatever other language is programmed. Thus Furby enjoys a virtual kind of growth in its communicative ability. Despite this capability, Furby gave almost no appearance of being intelligent, but it was widely perceived as being cute, and not only by children. (When my wife and I gave a party during the Furby craze, some of our friends brought their children along, but it was the adults who most monopolized our Furby.)

  While conducting their market research prior to designing robot pets, the most successful companies have discovered that artificial interactive pets sell better when they resemble real animals in appearance and behavior, when they simulate the experience of traditional pet ownership, thereby creating similarities that cause our perception of them to influence our emotional attachment to them. The more animal-like they are, the more attached we become. This is especially true with children, who will describe their feelings toward a pet robot in terms similar to those they employ when talking about their friends, a phenomenon known as transference.* This type of treatment of pet robots by children has been explored by Sherry Turkle’s group at MIT. They described how one of their child subjects, a soft-spoken, intelligent, and well-mannered ten-year-old girl named Melanie, reacted to the robotic dog AIBO† and the electronic doll My Real Baby‡:

  Melanie believes that AIBO and My Real Baby are sentient and have emotions. She thinks that when we brought the robotic dog and doll to her school “they were probably confused about who their mommies and daddies were because they were being handled by so many different people.” She thinks that AIBO probably does not know that he is at her particular school because the school is strange to him, but “almost certainly does know that he is outside of MIT and visiting another school.” She sees her role with the robots as straightforward; it is maternal.

  One of Melanie’s third-grade classmates treats My Real Baby as an object to explore and handles it very roughly, poking its eyes, pinching its skin to test its “rubber-ness” and putting her fingers roughly inside its mouth. Observing this, Melanie comes over to rescue the doll. She takes it in her arms and proceeds to play with it as though it were a baby, holding it close, whispering to it and caressing its face. When she is about to take it home, Melanie says, “I think that if I’m the first one to be with her then maybe if she goes home with another person [another study participant] she’ll cry a lot…because she doesn’t know, doesn’t think that this person is its mama.”6

  Although children are, in theory, more gullible than adults, they often represent the most fundamental of human reactions. Because of this their behavior and reactions can tell us a great deal about ourselves, enabling robot designers to learn about some of the basic needs in human companionship, and as robot manufacturers have learned more about what makes a robot attractive as a companion, so different robot applications have sprung up. The most popular and hence the bestselling robots have been those produced for entertainment: Sony’s robotic dog, AIBO; Honda’s walking android, ASIMO, which is even able to climb stairs; robots that can tell jokes; and the 2004 bestseller, Robosapien. Possibly because electronic learning aids were so popular during the late 1980s and much of the 1990s, educational robots have also proved to be a marketing success. Other robots of which prototypes have been demonstrated include Tohoku University’s ballroom-dancing androids that can predict the movements of a dancing partner, enabling these robots to follow a fellow dancer’s lead without stepping on any toes. Another example is the NEC Corporation’s personal robot that can recognize the faces of individual members of a family, entertain family members with its limited speech ability, and act as an interface to control the television and e-mail. A study of children aged three to five, by Thomas Draper and Wanda Clayton at Brigham Young University, found that robots with some sort of persona, robots who move and smile and say something in praise of a child’s success, make better teachers and engender a greater level of motivation in their pupils than do inanimate, machinelike robots that do not talk. The My Real Baby doll, manufactured by the toy giant Hasbro, is a good example—it speaks, it makes realistic sounds, its face moves in babyish ways and it exhibits several human emotions.

  The interactive aspect of a robot’s being is becoming an important or even an essential element of its usefulness. Carer robots and teacher robots are just two examples. As the learning abilities of robots develop from the primitive to the sophisticated, so robots will be able to adapt to the needs and desires of their human partners. No longer will it be necessary to redesign or even reprogram a robot to perform some new task for us; instead the robots of the future will learn by watching what makes us happy and grateful and will sense our desires and satisfy them. These artificially intelligent entities will no longer be perceived as some sort of machine. Rather they will become accepted as good companions. It is the leap into this realm of relationships capable of satisfying human needs that has spawned the new disciplines of robotic psychology and robotherapy.

  These new disciplines focus on the psychological aspects of our relationships with robots. While regular psychotherapists aim to help us gain some useful introspection into our own problems—both problems of self and those born out of our relationships with other human beings—robopsychology is concerned specifically with problems born of our relationships with robots. It is a highly complex minefield of new notions about relationships, in which the different ways people interact with robots and the different types of robot personality both have an effect. These effects can include caring and other t
herapeutic regimes, tailored by the robot designer (or, eventually, by the robot itself) to the specific needs of the individual.

  Carer robots for the elderly belong to a product category that is fast attracting the interest of major manufacturers, particularly in Japan. In 2004 a “robot suit” was launched for the elderly, a motorized, battery-operated pair of trousers designed to help the aged and infirm to move around on their own. Then there is the Wakamaru, a mobile, three-foot-high talking robot equipped with two camera eyes, used mainly by the Japanese to keep an eye on their elderly parents at home. Sanyo has even developed a robot for bathing and shampooing the elderly. According to the Japan Robot Association, products of this sort will increase Japanese sales of domestic robots to $14 billion in 2010 and $40 billion in 2025 because of the accompanying marked rise in the percentage of senior citizens in the population, a rise that has created huge interest in how best to satisfy the needs of the elderly. Similar bulges in the age statistics will soon hit just about every developed country, partly because people in those countries are living longer than their parents and grandparents did and partly because of the post-WWII baby boom. As a result, robots represent one of mankind’s best chances of being able to cope by providing therapeutic care for the aged.

  Research into the development of robot pets for therapeutic purposes has resulted in, among other technologies, artificial fur that incorporates touch sensors, allowing an artificial pet to respond when it is stroked. This touchy-feely attribute further increases the therapeutic value of a pet when combined with a robot’s lifelike appearance and behavior patterns. The pleasure of stroking a pet, together with the responses programmed into the pet for when it is stroked, have been found to enhance the experience for the elderly from both a psychological and a physiological perspective, thereby creating a friendlier mood in the patient. As a result, the moods of the patients and their overall feelings of comfort are generally improved by the stroking experience.