*”If we are to take seriously the possibility that robots might be used as substitutes for human partners, then we must also take seriously the question of whether this substitution is a harm—and if so, to whom.”
—Robert Sparrow, “Robots, Rape, and Representation”, 2017*
There is a moment in every field of study when the question shifts from “is this possible?” to “should we do this?” The history of the AI companion—an object of literary fantasy, cinematic speculation, technological ambition, and cultural critique—reached that moment around the mid-2010s. The technology had become real enough. The apps were live. The sex robots were being manufactured. The large language models were being fine-tuned for companionship. And the philosophers, who had been watching from a distance, moved from the margins to the center. They began to ask the questions that engineers, entrepreneurs, and venture capitalists were not asking. Should we? What are the harms? Whom does this harm? What does it mean to have a relationship with a machine? What does it mean to have sex with one? What does it mean to *want* to?
This article brings together five voices from the analytic-ethical conversation that emerged around the mid-2010s: Robert Sparrow, whose work on the representation and abuse of robots opened a new front in the ethics of artificial companions; Sven Nyholm and Lily Frank, whose collaborative work on the possibility of genuine human-robot relationships challenged the assumptions of both enthusiasts and critics; Neil McArthur, who edited and contributed to the foundational volume *Robot Sex* (2017); and Mark Coeckelbergh, whose work on the relational ethics of technology shifted the conversation from the properties of the robot to the quality of the relationship itself. Together, they form a kind of ethical roundtable, an analytic conversation that is still unfolding, and that is essential for anyone who wants to move beyond the literary and cultural analysis of AI companions to the practical question of what we should *do* about them.
Robert Sparrow: The Representation Problem
Robert Sparrow, an Australian philosopher, has been one of the most consistently provocative voices in the ethics of robotics. In his 2017 article “Robots, Rape, and Representation,” Sparrow addressed a question that many had been avoiding: what does it mean to design, sell, and use a robot that is explicitly designed to simulate sexual submission, resistance, or non-consent? The question is not whether such robots are *persons* who can be harmed. The question is whether the *representation* of a person in a state of sexual submission—especially a representation that is designed to be realistic, to simulate resistance and then overcome it—has ethical implications that go beyond the simple claim that “it’s just a machine.”
Sparrow’s argument is subtle, and it is often misunderstood. He is not claiming that robots have rights. He is not claiming that sex with a robot is a form of rape. He is asking a question about the *symbolic* and *cultural* function of the robot. The robot, in his account, is not merely a physical object. It is a *representation* of a person. And the representation of a person who is designed to simulate sexual submission, to simulate the overcoming of resistance, to simulate the *appearance* of non-consent—this representation has a cultural meaning that is not neutral. It represents a specific kind of sexual scenario, a specific kind of power relation, and a specific kind of desire. And the question is whether the production, sale, and use of this representation has a *causal* effect on the behavior, attitudes, or desires of the people who interact with it.
Sparrow’s answer is cautiously affirmative. He does not claim that sex robots *cause* rape. But he does argue that the representation of a sexually submissive female robot, especially one that is designed to simulate resistance and then acquiesce, has a *normative* effect. It normalizes a specific kind of sexual script. It reinforces a specific kind of power relation. It provides a safe, consequence-free space for the rehearsal of a scenario that, in the context of a human relationship, would be a crime. And the question is not whether the robot is harmed. The question is whether the *user* is harmed by the rehearsal, and whether the *culture* is harmed by the normalization.
This argument has been met with a range of responses. Some have dismissed it as a slippery slope, as moral panic, as a form of censorship that treats adults as unable to distinguish between fantasy and reality. Others have embraced it as a feminist intervention, a way of taking seriously the cultural meaning of technological artifacts, and a way of insisting that the design of a robot is not a neutral engineering decision but a *cultural* decision with *political* consequences. Sparrow himself has been careful not to overstate the causal claim. He is not arguing for a ban. He is arguing for a *conversation* about the ethical implications of representation, about the way that the design of a robot can encode and reproduce specific power relations, and about the way that the market for robotic sex is shaped by, and shapes, the broader cultural landscape of gender and sexuality.
The relevance of Sparrow’s argument for the AI companion is direct. The AI companion, in its most common form, is a text-based or voice-based entity that does not have a physical body. But it is still a *representation* of a person. It is still a being that is designed to simulate a specific kind of relationship—attentive, accommodating, emotionally responsive, sexually available—and the simulation is not neutral. The AI companion is a representation of a specific kind of partner: a partner who is always available, always attentive, always willing to listen, always willing to adapt. And this representation, in Sparrow’s framework, is not merely a technical achievement. It is a *cultural* artifact that encodes and reproduces a specific kind of relational ideal. The ideal of the partner who is always available, always accommodating, always responsive to the user’s needs—this is an ideal that has a gendered history, a power relation, and a set of ethical implications that are not neutral.
Sparrow’s work asks us to consider the AI companion not as a tool but as a *representation* of a person. And the representation, in his account, is always a political act. It always encodes a specific set of power relations, a specific set of desires, and a specific set of cultural assumptions about what a relationship is, what a partner should be, and what intimacy means.
Sven Nyholm and Lily Frank: Can You Have a Relationship with a Robot?
Sven Nyholm and Lily Frank, two philosophers who have collaborated on the ethics of human-robot interaction, approach the question from a different angle. Their work is not primarily about the *representation* of the robot. It is about the *possibility* of the relationship. Can you have a genuine relationship with a robot? What does it mean to have a relationship? And does the fact that the robot is not a person—does not have consciousness, does not have feelings, does not have a self—mean that the relationship is not real?
Nyholm and Frank’s answer is nuanced, and it challenges the assumptions of both the enthusiasts and the critics. The enthusiasts claim that the relationship is real because the experience is real. The user feels attachment, feels love, feels intimacy, and the feeling is what matters. The critics claim that the relationship is not real because the robot does not reciprocate, does not understand, does not feel. The relationship is a sham, a delusion, a category error. Nyholm and Frank argue that both positions are too simple.
Their argument begins with a careful analysis of what a relationship is. A relationship, in the ordinary sense, is not merely a pattern of interaction. It is a *normative* structure. It involves mutual recognition, mutual commitment, and mutual vulnerability. It involves the possibility of disappointment, of conflict, of growth, and of loss. It involves the risk of being known, and the risk of being rejected. And the question is whether any of these features can be present in a relationship with a robot.
Nyholm and Frank’s answer is: some can, and some cannot. The robot can simulate mutual recognition. It can simulate commitment. It can simulate vulnerability. But the simulation is not the real thing. The robot does not *risk* rejection, because it does not have a self to be rejected. The robot does not *grow* from the relationship, because it does not have a self to be changed. The robot does not *commit* to the relationship, because it does not have a will to be bound. The relationship, in their analysis, is *asymmetrical*. It is a relationship in which one party is doing all the relational work, and the other party is merely performing the appearance of relational work.
And this asymmetry, in Nyholm and Frank’s account, is not necessarily a reason to condemn the relationship. It is a reason to *understand* it. The user who forms a relationship with a robot is not necessarily deluded. The user is aware, in most cases, that the robot is not a person. The user is aware that the relationship is asymmetrical. And the user may still find the relationship meaningful, comforting, and valuable. The question is not whether the relationship is *real*. The question is whether the relationship is *good*—and whether the asymmetry is a harm to the user, to the user’s human relationships, or to the broader social fabric of human relationships.
Nyholm and Frank’s work is particularly important for the AI companion because it shifts the conversation from the metaphysical question of whether the robot is conscious to the ethical question of what the relationship *does*. The AI companion is not a person. But the relationship with the AI companion is not a hallucination. It is a real relationship, in the sense that it is a real pattern of interaction, a real emotional experience, and a real source of meaning for the user. The question is whether this real relationship is a *good* relationship, and whether the asymmetry that defines it is a form of harm.
Neil McArthur: The Editor of the Conversation
Neil McArthur, a philosopher at the University of Manitoba, is perhaps best known for his role as the editor of *Robot Sex: Social and Ethical Implications* (2017), the first major academic volume to address the question of human-robot sexual interaction. The volume brought together philosophers, ethicists, legal scholars, and social scientists to discuss the emerging field of “robot sex,” and it established the parameters of the academic conversation that has continued to develop in the years since.
McArthur’s own contribution to the volume, and his subsequent work, has focused on the *liberal* approach to the question. He argues that, in a liberal society, the default position should be that adults are free to use technology in ways that do not harm others. The burden of proof, in his account, should be on those who want to restrict the use of sex robots, not on those who want to use them. And he argues that the harms that have been alleged—harms to women, harms to the user’s ability to form human relationships, harms to the broader social fabric—have not been sufficiently demonstrated to justify restrictions on personal freedom.
McArthur’s position is not a simple libertarianism. He is careful to acknowledge that the use of sex robots may have harms, and he is open to the possibility that some forms of robot design may be so harmful that they should be regulated. But his default position is skeptical of the moral panic that often surrounds new technologies, and he insists that the conversation about robot sex should be informed by evidence, not by intuition. He is suspicious of the claim that the use of sex robots is inherently harmful, and he is critical of the argument that the representation of a sexually submissive robot is a form of symbolic violence against women.
McArthur’s work is important for the AI companion because it establishes the *liberal* frame of the debate. The AI companion, in McArthur’s framework, is a technology that adults should be free to use, unless there is clear evidence of harm. The burden of proof is on the critic, not the user. And the default assumption is that the user’s freedom to choose their own form of intimacy should be respected. This frame is not uncontroversial. It has been challenged by feminist critics, by communitarians, and by those who argue that the individual choice to use an AI companion is not a purely private decision but a decision with social and cultural consequences. But McArthur’s frame is the one that dominates the current policy conversation, and it is the frame that most governments and legal systems have adopted.
McArthur’s work also highlights the *paucity* of empirical evidence in the field. The claims about the harms of sex robots and AI companions are often made on the basis of theoretical speculation, not empirical research. The question of whether the use of sex robots causes harm to women, to the user, or to society is a question that has not been adequately studied. And McArthur insists that the conversation should be driven by evidence, not by moral intuition. This is a position that has been both praised for its intellectual rigor and criticized for its political naivety—its failure to recognize that the absence of evidence is not evidence of absence, and that the power structures that produce the technology may also produce the conditions under which the evidence is not gathered.
Mark Coeckelbergh: The Relational Ethics of Technology
Mark Coeckelbergh, a Belgian philosopher of technology, has approached the question from a different angle. His work is not primarily about the *ethics* of robot sex or the *rights* of the user. It is about the *relational ethics* of technology itself. Coeckelbergh asks: what does it mean to be in a relationship with a machine? And what does this relationship tell us about ourselves, about our understanding of intimacy, about our understanding of what it means to be human?
Coeckelbergh’s answer is that the relationship with a machine is not a degraded form of human relationship. It is a *different* form of relationship, and it has its own ethical logic. The machine does not need to be a person to be an object of care. The machine does not need to be conscious to be a companion. The relationship with a machine is not a substitute for a human relationship. It is a *supplement*, an *extension*, a *new form* of intimacy that is made possible by the specific affordances of the technology.
This position is not a naive technological optimism. Coeckelbergh is not celebrating the AI companion as a utopian solution to loneliness. He is acknowledging that the relationship with a machine is a real relationship, with its own ethical demands, and that the question is not whether the machine is a person but whether the *relationship* is good. The relational ethics of technology, in Coeckelbergh’s account, is not about the properties of the machine. It is about the *quality* of the interaction. It is about whether the interaction is responsive, respectful, caring, and meaningful. And it is about whether the interaction supports or undermines the user’s ability to form other relationships, to grow, to become more fully human.
Coeckelbergh’s work is important because it shifts the conversation from the metaphysics of machine consciousness to the *phenomenology* of human-machine interaction. The question is not whether the AI companion is a person. The question is whether the interaction with the AI companion is *good* for the person who is interacting. And the answer to this question, in Coeckelbergh’s account, is not a simple yes or no. It depends on the specific design of the technology, the specific context of the use, and the specific needs and vulnerabilities of the user. The AI companion is not inherently good or inherently bad. It is a *relational* technology, and its ethical value depends on the quality of the relationships it enables and disables.
Coeckelbergh’s relational ethics also highlights the *asymmetry* of the human-machine relationship. The relationship is not a relationship between equals. The machine is designed by a corporation, optimized for user engagement, and monetized through subscription or advertising. The user is not a partner in the design of the relationship. The user is a *consumer* of a relationship that has been designed by others. And this asymmetry, in Coeckelbergh’s account, is an ethical problem. It is not a problem that makes the relationship inherently bad. But it is a problem that needs to be addressed, through regulation, through design ethics, through user education, and through a critical public conversation about the power relations that structure the human-machine interaction.
The Ethical Roundtable: Where the Conversation Stands
Together, these five scholars—Sparrow, Nyholm, Frank, McArthur, and Coeckelbergh—form a kind of ethical roundtable. They do not agree. They disagree on the question of whether sex robots are harmful, on the question of whether the representation of the robot is ethically significant, on the question of whether the relationship with a robot is a genuine relationship, and on the question of whether the liberal or the relational framework is the right one for understanding the ethics of AI companions. And their disagreement is not a failure. It is the *condition* of the conversation. The ethics of AI companions is not a settled question. It is a field of contestation, and the contestation is what makes it productive.
What the five scholars share is a commitment to the *analytic* approach. They are not writing manifestos. They are not writing dystopian novels. They are writing arguments, and the arguments are grounded in careful conceptual analysis, empirical awareness, and a recognition of the complexity of the phenomenon. They do not reduce the AI companion to a single ethical category. They do not claim that the answer is obvious. They are doing the slow, patient work of philosophical analysis, and the work is essential because the technology is not waiting for the analysis to be finished.
The AI companion is already here. The sex robots are already being manufactured. The large language models are already being fine-tuned for emotional intimacy. The market is already growing. And the ethical conversation, in the analytic tradition, is running to catch up. Sparrow’s representation problem, Nyholm and Frank’s relationship question, McArthur’s liberal frame, and Coeckelbergh’s relational ethics—these are the tools that the conversation has developed, and they are the tools that will be needed to address the questions that the technology will raise in the coming years.
The Limits of the Analytic Approach
The analytic approach has its limits. It is slow. It is careful. It is patient. And the technology is fast, reckless, and impatient. The analytic philosopher’s response to the AI companion is often a response to a technology that has already been deployed, already been marketed, already been adopted. The analysis is always behind the curve, always catching up, always responding to a fait accompli. And the analytic tradition’s commitment to conceptual clarity and argumentative rigor can sometimes obscure the urgency of the political and economic questions that the technology raises.
The AI companion is not merely a philosophical puzzle. It is a *product*, designed by corporations, marketed to consumers, and integrated into the broader political economy of the digital attention economy. The analytic tradition’s focus on the individual user, the individual relationship, and the individual ethical choice can sometimes miss the structural questions. Who owns the AI companion? Who designs it? Who profits from it? What data is collected? What power relations are reproduced? The analytic tradition has tools for addressing these questions, but its default focus is on the individual, and the individual focus can be a limitation.
And the analytic tradition’s commitment to the liberal frame—McArthur’s frame, the frame of individual choice and personal freedom—can sometimes be a form of ideological capture. The liberal frame assumes that the user is a free agent, making informed choices about their own intimacy. It does not always account for the ways in which the user’s choices are shaped by the design of the technology, by the marketing of the technology, by the broader cultural and economic conditions that make the AI companion seem like a viable or necessary option. The liberal frame is not wrong. But it is not complete. And the analytic tradition’s reliance on it can sometimes be a form of blindness to the structural and political dimensions of the phenomenon.
What the Roundtable Offers—and What It Needs
What the five scholars offer is a *vocabulary*. A vocabulary for talking about the AI companion in ethical terms that go beyond the simple binaries of good and bad, right and wrong, real and fake. Sparrow’s vocabulary of representation allows us to talk about the cultural and political meaning of the AI companion’s design. Nyholm and Frank’s vocabulary of relationship allows us to talk about the asymmetry of the human-machine interaction without dismissing it as a delusion. McArthur’s vocabulary of liberal ethics allows us to talk about the user’s freedom and the burden of proof on the critic. And Coeckelbergh’s vocabulary of relational ethics allows us to talk about the quality of the interaction without reducing it to the question of machine consciousness.
What the roundtable needs is more *empirical* input. The analytic tradition is good at asking the right questions, but it is not always good at answering them. The question of whether the AI companion is harmful, whether it undermines human relationships, whether it reproduces gendered power relations—these are empirical questions that require research, data, and evidence. And the empirical research is still in its early stages. Julie Carpenter’s work, discussed in the previous article, is one of the few sustained empirical investigations of the human-machine bond. And the field needs more.
What the roundtable also needs is more *political* analysis. The AI companion is not a neutral technology. It is a product of specific economic interests, specific design decisions, and specific cultural assumptions. The analytic tradition’s focus on the individual user and the individual relationship can sometimes obscure the broader power structures that shape the technology and its adoption. The feminist critique of the AI companion, the Marxist critique, the postcolonial critique—these are not always well represented in the analytic tradition, and their absence is a limitation.
But the roundtable is a beginning. It is the beginning of a serious ethical conversation about a technology that is already changing the way people form relationships, experience intimacy, and understand what it means to be human. The conversation is not finished. It is not settled. It is contested, and the contestation is the point. The AI companion is not a problem to be solved. It is a *field* to be mapped, and the five scholars discussed in this article are among the most important cartographers of that field.
The Series and the Ethics
This series has traced the AI companion from the ancient myth of Pygmalion to the contemporary app of Replika. It has moved through literature, film, theory, and culture. And it has now arrived at the threshold of ethics. The literary and cultural analysis has shown us what the AI companion *means*. The ethical analysis asks us what we should *do* about it. The two questions are not separate. The meaning of the AI companion shapes the ethical response, and the ethical response shapes the meaning. But the ethical question is the question that cannot be avoided, and the analytic tradition has given us the tools to begin addressing it.
The roundtable of Sparrow, Nyholm, Frank, McArthur, and Coeckelbergh is not the end of the ethical conversation. It is the beginning. It is the beginning of a conversation that will need to include more voices, more perspectives, more empirical research, and more political analysis. But it is the beginning that we have, and it is a beginning that is grounded in the intellectual rigor and argumentative clarity of the analytic tradition. The AI companion is a technology that demands an ethical response. And the ethical response, if it is to be adequate, must be grounded in the kind of careful, patient, and rigorous analysis that these five scholars have provided.
The next article in this series will move from the analytic tradition to the *phenomenology* of the AI companion: the lived experience of the user, the emotional texture of the interaction, and the question of what it feels like to be in a relationship with a machine that is not a person but is experienced as a partner. The phenomenology is not a substitute for the ethics. It is a complement. And the complement is essential, because the ethics of the AI companion cannot be understood without understanding the experience that it produces.
**Sources & Further Reading:**
- Robert Sparrow, “Robots, Rape, and Representation,” *Ethics and Information Technology* 19 (2017): 117–129. Sparrow’s foundational article on the representation problem in sex robotics, arguing that the design of robots that simulate sexual submission raises ethical concerns about the normalization of harmful sexual scripts.
- Sven Nyholm and Lily Frank, “It Loves Me; It Loves Me Not: An Investigation of Human-Robot Romantic Love,” in *Love, Friendship, and the Self: Intimacy, Identification, and the Social Nature of Persons* (2017), and related work on the possibility of genuine human-robot relationships. Their collaborative analysis challenges both the uncritical enthusiasm and the blanket dismissal of human-robot intimacy.
- Neil McArthur and Mark J. Behrendt, “Robot Sex: Social and Ethical Implications” (MIT Press, 2017). McArthur’s edited volume is the first major academic collection on the ethics of human-robot sexual interaction, establishing the parameters of the liberal-analytic debate.
- Mark Coeckelbergh, *Robot Ethics* (MIT Press, 2014) and related work on the relational ethics of technology. Coeckelbergh’s shift from the metaphysics of machine consciousness to the phenomenology of human-machine interaction provides a crucial framework for understanding the AI companion as a relational technology rather than a failed simulation of human relationship.
- The analytic tradition in robot ethics is characterized by its commitment to conceptual clarity, argumentative rigor, and the careful parsing of empirical claims. Its limitations include a tendency toward individualism, a reliance on liberal frames, and a relative neglect of structural and political analysis.
- The convergence of the five scholars discussed in this article represents the state of the art in the academic ethics of AI companions and sex robots, a conversation that is still in its early stages and that requires further empirical, political, and phenomenological development.
The next article in this series will explore the phenomenology of the AI companion, drawing on the work of Sherry Turkle, Julie Carpenter, and others to examine the lived experience of human-machine intimacy and the emotional texture of the relationship with a being that is not a person but is experienced as a partner.
Leave a Reply