I. A new kind of object: From Rorschach to Relationship

I have studied the effects of computational objects on human developmental psychology for over twenty years, documenting the ways that computation and its metaphors have influenced our thinking about such matters as how the mind works, what it means to be intelligent, and what is special about being human. Now, I believe that a new kind of computational object — the relational artifact — is provoking striking new changes in the narrative of human development, especially in the way people think about life, and about what kind of relationships it is appropriate to have with a machine. Relational artifacts are computational objects designed to recognize and respond to the affective states of human beings–and indeed, to present themselves as having "affective" states of their own. They include children's playthings (such as Furbies and Tamagotchis), digital dolls that double as health monitoring systems for the homebound elderly (Matsushita's forthcoming Tama), sentient robots whose knowledge and personalities change through their interactions with humans, as well as software that responds to its users’ emotional states and responds with "emotional states" of their own.

Over the past twenty years, I have often used the metaphor of "computer as Rorschach" to describe the relationship between people and their machines. I found computers used as a projective screen for other concerns, a mirror of mind and self. But today’s relational artifacts make the Rorschach metaphor far less useful than before. These artifacts do not so much invite projection as they demand engagement. The computational object is no longer affectively "neutral." People are learning to interact with computers through conversation and gesture, people are learning that to relate successfully to a computer you do not have to know how it works, but to take it "at interface value," that is to assess its emotional "state," much as you would if you were relating to another person. Through their experiences with virtual pets and digital dolls (Tamagotchi, Furby, Amazing Ally), a generation of children are learning that some objects require (and promise) emotional nurturance. Adults, too, are encountering technology that attempts to meet their desire for personalized advice, care and companionship (help wizards, intelligent agents, AIBO, Matsushita's forthcoming Tama).

These are only the earliest, crude examples of the relational technologies that will become part of our everyday lives in the next century. There is every indication that the future of computational technology will include ubiquitous relational artifacts that have feelings, life cycles, moods, that reminisce, and have a sense of humor, which say they love us, and expect us to love them back. What will it mean to a person when their primary daily companion comes is a robotic dog? Or their health care "worker" is a robot cat? Or their software program attends to their emotional states and, in turn, has its own?. We need to know how these new artifacts affect people’s way of thinking about themselves, human identity, and what makes people special. These artifacts also raise significant new questions about how children apporach the question of ‘What is alive?" In the proposed research the question is not what the computer will be like in the future, but what will we be like, what kind of people are we becoming?

Relational artifacts are changing the narrative of human development, including how we understand such "human" qualities as emotion, love, and care. The dynamic between a person and an emotionally interactive, evolving, caring machine object is not the same as the relationship one might have with another person, or a pet, or a cherished inanimate object.

We have spent a large amount of social resources trying to build these artifacts; now it is time to study what is happening to all of us as we go forth into a world "peopled" with a kind of object we have never experienced before. We need to more deeply understand the nature and implications of this new sort of relationship — and its potential to fundamentally change our understanding of what it means to be human.

We need to be asking several kinds of new questions:

• How are we to conceptualize the nature of our attachments to interactive robots, affective computers, and digital pets?

• How does interacting with relational artifacts affect people’s way of thinking about themselves and others, their sense of human identity and relationships? How do the models of development and values embedded in the design of relational artifacts both reflect and influence our ways of thinking about people?

• What roles — both productive and problematic — can relational artifacts play in fulfilling a basic human need for relationship? Their first generation is being predominantly marketed to children and the elderly. What does this reflect about our cultural values about these groups? How will these objects influence their understanding of who they are as individuals and in the world? Are we reinforcing their marginality? Are we tacitly acknowledging that we do not have enough "human" time to spend with them?

II.

In the 1960s through the 1980s, researchers in artificial intelligence took part in what we might call the classical "great AI debates" where the central question was whether machines could be "really" intelligent. This classical debate was essentialist; the new relational objects tend to enable researchers and their public to sidestep such arguments about what is inherent in the computer. Instead, the new objects depend on what people attribute to them; they shift the focus to what the objects evoke in us. When we are asked to care for an object (the robot Kismet, the plaything Furby), when the cared-for object thrives and offers us its attention and concern, people are moved to experience that object as intelligent. Beyond this, they feel a connection to it. So the question here is not to enter a debate about whether relational objects "really" have emotions, but to reflect on a series of issues having to do with what relational artifacts evoke in the user.

In my preliminary research on children and Furbies, I have found that children describe these new toys as "sort of alive" because of the quality of their emotional attachments to the Furbies and because of their fantasies about the idea that the Furby might be emotionally attached to them. So, for example, when I ask the question, "Do you think the Furby is alive?" children answer not in terms of what the Furby can do, but how they feel about the Furby and how the Furby might feel about them.

Here, the computational object functions not only as an evocative model of mind, but as a kindred other. With these new objects, children (and adults) not only reflect on how their own mental and physical processes are analogous to the machine’s, but perceive and relate to the machine as an autonomous and "almost alive" self.

My work with children and computational objects has evolved into a decades-long narrative about the way computation has affected the way we make sense of the world. In many ways, the behaviors and comments of children have foreshadowed the reactions of adults. In the first generation of computer culture I studied, the children of the late 1970s and early 1980s tended to resolve metaphysical conflicts about machine "aliveness" by developing a concept of "the psychological machine"–concluding that psychology and a kind of consciousness were possible in objects they knew were not alive. This way of coping with the conundrums posed by computational objects was pioneered by children, and later adopted by adults. Later cohorts of children's responses to computational objects that were more complex and problematic in new ways again reliably foreshadowed the conclusions the culture at large would soon reach. First, they explained that although machines might be psychological in the cognitive sense (they might be intelligent, they might have intentionality), they were not psychological in the emotional sense, because they did not know pain, or love, they were not mortal, and they did not have souls. Soon after, the children I interviewed began consistently citing biology and embodiment as the crucial criteria that separated people from machines; they insisted that qualities like breathing, having blood, being born, and, as one put it, "having real skin," were the true signs of life. Now, I have begun to see a new pattern – children describe relational artifacts not as "alive" or "not alive", but as "sort-of-alive." Categories such as "aliveness" and "emotion" seem poised to split in the same way that the categories of "psychological" and "intelligent" did twenty years ago.

Children's reactions to the presence of "smart machines" have fallen into discernable patterns over the past twenty years. Adults' reactions, too, have been changing over time, often closely following those of the children. To a certain extent, we can look to children to see what we are starting to think ourselves. However, in the case of relational artifacts, there is more to the choice of children as subjects than a simple desire to stay ahead of the curve in anticipating changes in computer culture. By accepting a new category of relationship, with entities that they recognize as "sort-of-alive", or "alive in a different, but legitimate way," today's children will redefine the scope and shape of the playing field for social relations in the future. Because they are the first generation to grow up with this new paradigm, it is essential that we observe and document their experiences.

SHERRY TURKLE is a professor of the sociology of sciences at MIT. She is the author of The Second Self: Computers and the Human Spirit; Psychoanalytic Politics: Jacques Lacan and Freud�s French Revolution, and Life on the Screen: Identity in the Age of the Internet..

LINKS: Sherry Turkle's Home Page; See "The Cyberanalyst", Chapter 31 in Digerati