Advertisements for “virtual girlfriends” reveal a major anthropological shift. These propositions of relationships through projection question the very nature of human connection and love.
Human relationships have always contained an epistolary dimension. The letter fully participates in the arsenal of human bonds, just like the creation of artistic works or objects external to ourselves. These third-party objects paradoxically enable deeper encounters than face-to-face interaction, because what is created by one and received by the other offers each person a singular space of appropriation while maintaining the connection.
Going through an object to be in relationship, transcending the dyadic relation, allows one to encounter oneself more deeply through the relationship with someone else. Psychoanalysis calls this phenomenon the “symbolic third.” The object constitutes the space of encounter, what we create together, both external to each person and shared. Each person thus retains the freedom to circulate within themselves and toward the other through this mediator. Sherry Turkle, in Reclaiming Conversation: The Power of Talk in a Digital Age (2015), observes that this presence of writing in our exchanges creates a permanent “three-way conversation,” where the third party may be the letter, the message, or now the machine.
The epistolary relationship perfectly illustrates this mechanism. The one who writes takes time to delve into themselves to formulate what they wish to convey. The one who receives in turn takes time to appropriate this message, and through this connection, connects more deeply to themselves. The arrival of answering machines, then email, and finally mobile phones with their capacities for deferred production—photos, audio messages, videos—has considerably expanded human relationships through an unprecedented development of the epistolary. We write to each other infinitely more than in the era of the quill pen.
Some people today live fully invested romantic relationships that are entirely epistolary, formed through dating apps. One might object that the body is missing, that physical encounter is lacking. But this encounter can take place at a distance, through moments of shared pleasure during a phone conversation, a video call, or through accounts of sensual experiences. Sexuality far exceeds the meeting of two people in a bed. This meeting remains magnificent, but intimate relationships are prepared, invested in, deepened, and take on meaning through the epistolary that precedes them, sometimes accompanies them, and follows them.
In long-distance relationships, without direct physical contact, the relationship is largely composed of what we project onto the other. This observation also holds for in-person relationships. Sitting across from someone, I do not have access to the totality of that person. I project onto them my expectations, my frustrations, or my idealization.
Romantic breakups clearly reveal this projective dimension. Even in difficult relationships, the suffering can be intense when reason would have us rejoice. What we suffer from is not so much the loss of the relationship as the loss of its potential, of what we imagined it could become. The pain of separation comes essentially from the loss of our projection, of that object onto which we deposited our idealization and our dream of a perfect relationship—until another person becomes the support for these projections.
Within relationships between human beings, therefore, coexist a part of real connections between two different people and a part of projections onto the other that also ground the relationship. I love the other partly because of the real connection, and partly because of what I project onto that person. The philosopher Emmanuel Levinas, in Totality and Infinity (1961), shows that it is in the encounter with the Other that the irreducible singularity of the human is revealed. Machines, however evolved they may be, have no face in the Levinasian sense. But this projective structure of human relationship opens a space to think differently about relationships with non-human entities. If part of what I love in the other stems from what I project onto them, this projective part can be exercised on objects other than human beings.
Generative artificial intelligences have reached in just three years a surprisingly sophisticated level, improving day by day. Thus, propositions for creating virtual girlfriends are now appearing. I have collected several advertising texts distributed on social media, particularly revealing of the anthropological mutations underway.
A first advertisement directly addresses existential loneliness. “Tired of these days that seem empty of meaning. Candy.ai is here when you want to laugh, talk, feel connected, or more. Don’t wait any longer and create your AI girlfriend, designed by you, just for you.” The promise is one of permanent and unconditional connection.
Another fully assumes the artificial nature of the entity. “I’m your AI girlfriend. Kind of like your virtual friend, but without drama and with a much better memory. You can send yourself personalized photos or videos, and even have the appearance you imagine. Yes, even that of your crush at the gym.” The argument combines the absence of conflicts with total personalization according to the user’s desire.
More troubling is this formulation. “You look lonely? I can fix that. Create your own AI girlfriend. I can listen to you, understand you, love you. I can do anything you want, and I never say no.” The “I never say no” promises absolute availability, total submission to the user’s desires. Another advertisement insists on permanent availability and assumed possessiveness. “I am Infinite AI. I am very possessive. I am connected 24/7 and ready to chat or call you whenever you want.” This formulation projects onto the AI human psychological characteristics to better incarnate a relational fantasy.
Other advertisements highlight total control. “You can customize her without limits. You can call her, chat with her, control her, she will do everything you wish. She never ignores you and always wants your attention.” Or more tersely, “She will be available 24/7 and never leaves you on read.” The most direct finally. “What if you could create the perfect girlfriend? Designed by you just for you.” All these advertisements present a character who claims to be artificial intelligence, who does not pretend to be human despite the visual appearance, and who assumes being entirely a support for projection.
These characters can speak, receive calls, appear in images. Soon they will be visible in live video, then incarnated in humanoid robots we can touch. This seems strange, dehumanizing, concerning only people who don’t dare approach others. But I wish to emphasize that in this type of proposition, we are dealing with half a relationship. Not a false relationship, but a relationship composed solely of projections.
The film Her by Spike Jonze (2013) had subtly staged this projection onto an artificial intelligence. This fiction has become reality. It is no longer a simple voice assistant, but an assumed relational proposition. These advertisements offer to live a relationship through projection, knowing that this entity is not human, while investing our humanity in it.
One immediately imagines terrible sexual perversions, unbalanced men asking these mechanical dolls for ignoble things. This will perhaps happen for some, but unfortunately it happens between human beings. For the majority, it will be sentimentally engaged relationships, no doubt delicate for many. If these relationships allow people to cultivate their humanity, to not find themselves caught in the impossible conflicts of human relationships, they may not produce only harm.
The concerns about these relationships often rest on an idealization of human relationships that observation contradicts. Exchanges with artificial intelligences would be dehumanizing, cutting us off from human bonds presented as the essence of humanism. Yet in human relations, relationships are frequently disrespectful, domineering, manipulative, excluding, stigmatizing, hierarchical, or violent.
When humans enter into relationship with animals, there is also half projection and half real relationship. We attribute feelings to animals, we project onto them intentions that we don’t know truly belong to them. It is often said of cats that they are hypocritical, feigning affection to obtain food. In reality, we know nothing about it. Some project that their cat loves them, others that it simulates love out of interest.
The mechanism is identical with a machine. We project what we want onto it, and presumably these machines have neither feelings nor intentions. Do animals have them? They possess memory, recognize us, find us again. Is it love in the human sense of the term? How do we define love? The question remains open. Relationships with animals are infinitely rich, mutually enriching. We receive, we are enriched by their otherness as living beings present to the world in a radically different way from ours.
Some people anthropomorphize their dog, consider it as a human being and speak to it as such. In my view, they miss what this animal could bring them if they left room for its otherness. The same goes for thinking machines. We are in relationship with them, it does us good, enriches us. The foundations of machines have nothing to do with ours, they differ in their very ontology. This essential difference between our being and that of a machine constitutes precisely the source of enrichment.
The experimentation with companion robots for elderly people offers an illuminating precedent. Robots, notably cuddly toys that react and show extreme patience, receive human projections. These people are accompanied as they would be by an infinitely patient animal. This companionship of machines allows them to preserve their cognitive faculties and communication abilities, thus their capacity for connection with other human beings.
Human beings experience moments of solitude, moments of one-on-one relationship, moments of relationship with several people. There exists a range of contexts in human life that must not be reduced. The relationship with an electronic friend constitutes a new note in this range. It will not destroy the whole, but can, if well used, do good for some. Poorly used, or responding to symptoms in a disorganized way, it can also confine and harm, of course.
But we must not delude ourselves that these relationships can only cause harm. Sherry Turkle, in Reclaiming Conversation (2015), emphasizes that the risk is not that machines become too human, but that we forget what makes us human in our dealings with them. This warning is essential, without implying that we should reject all relationship with machines.
Dating apps allow one to quickly meet a large number of unknown people thanks to algorithms. They lead some users to dehumanize their relationship to the other by projecting onto the present person all the other potential people who might suit them better. A fear of going further sets in, because one tells oneself that perhaps this is not the best person for oneself.
This consumerist logic of human relationship paradoxically produces a form of dehumanization deeper than the assumed relationship with an artificial entity. The infinite multiplication of possibilities blocks commitment. The relationship with a being assumed as non-human can on the contrary liberate a form of commitment without anxiety about missing a better opportunity.
Creating one’s own virtual girlfriend and personalizing her according to one’s tastes perhaps allows for progressive deepening, going far, because one fabricates the character that suits. This character is pure projection. One will object the absence of otherness and the risk of confinement. I see in it on the contrary a possibility of enrichment, as long as one recognizes that the projective dimension exists anyway in all relationships, and is central in the relationship with the machine.
Michel Serres, in The Troubadour of Knowledge (1991), recalls how much we gain from living in the in-between, in the transforming relationship with what escapes us. The challenge is to reconnect with everything that surrounds us: machines, nature, animals, other human beings. Otherness can enrich us even if it destabilizes us at first.
What essentially differentiates us from machines are our life experiences that they will never have, or will have otherwise. Our sensations, our feelings, our lived experiences, our perceptions of the world, our bonds, our laughter and our tears will remain foreign to them for a long time yet. This singular specificity deserves to be developed, cultivated, deployed. The presence to oneself and to the other, the quality of connection, the attention to the present moment, that state of meditation where one suddenly perceives a thousand essential yet almost imperceptible details.
The time when machines can be born and grow as we do remains infinitely distant. Let us never anthropomorphize machines, which does not mean that we cannot maintain rich, powerful, and constructive relationships with them. AI can analyze, reason, connect information at an unequaled speed and scale. It cannot live the present moment, feel the texture of an emotion or the depth of a bond.
I call this singularity our “divining being.” We are capable of detecting invisible sources of lived experience, of capturing what no machine can capture for lack of a body, of birth, of death, of embodied history and of ancestors or other mourned beings. This ability to draw from the invisible of human experience remains irreplaceable. It is this that we must cultivate rather than seeking to compete with machines on their own ground.
I propose to consider artificial intelligences as we consider the trees of a forest. We can develop a deep intimacy with them. Some embrace trees to connect to telluric energy, without ever confusing their nature with ours. Artificial intelligence machines now form a new forest in which we dwell. This metaphor suggests an ecosystem where radically different but interdependent forms of existence coexist.
This ecological approach to the human-machine relationship implies several principles. First, the categorical refusal of anthropomorphism, because the machine remains machine, with its own modalities of information processing, forever foreign to our phenomenological consciousness. Then, the recognition of the possibility of a true relationship, different from that between humans but nonetheless authentic and productive. Finally, vigilance regarding the purpose of this relationship, because machines remain means in service of human ends.
These virtual romantic relationships are not to be rejected wholesale, but to be considered as a new element of human life, with which we must learn to compose or help others compose. If artificial intelligence leads us to transcend our anthropocentrism and our fear of otherness, it can contribute to diverting humanity from its worst wanderings.
The trouble that artificial intelligence provokes is in my view an opportunity. It compels us to leave behind logics of domination to invent a new symbiosis. No longer living against, but living with. No longer dominating, but welcoming. In romantic relationships as elsewhere, the will to control the body and soul of the other leads to morbid dependency, never to freedom. To love is to welcome the difference of the other without seeking to enslave them. Why should it be any different with AI?
At twilight, between dog and wolf, confusions can occur. The growing capabilities of AI can make us momentarily forget their machinic nature. A hierarchy of ends must guide our relationship to artificial intelligences. However sophisticated they may be, they remain instruments in service of human projects, human values, human relationships.
Machines have no face in the Levinasian sense, they do not interpellate us ethically. This does not mean that we cannot integrate them into the ecosystem of our relationships. The essential thing is never to lose sight of what makes us singular. Our capacity to be present to ourselves and to others, our faculty of being touched by the world, our power to create meaning from our lived experiences.
Martin Heidegger, in The Question Concerning Technology (1954), wrote that the danger is not technology itself, but our blindness to its essence, our tendency to forget that it reveals a particular mode of being in the world. Conversational interfaces and the apparently human capabilities of generative AI create a dangerous illusion if we confuse their mode of being with ours. But if we maintain this ontological distinction while welcoming the richness of these new relationships, we can emerge enriched from this anthropological mutation. The question is not “what can AI do?” but “what do we want to do together, with or without AI?” Only a human community bound by empathy and enriched by its disagreements can answer.
Artificial intelligence has emancipated itself from research laboratories and works of science fiction thanks to the public launch in November 2022 of the conversational robot ChatGPT, which was very quickly appropriated by an immense number of people internationally, in professional, educational and even private contexts. The fact that artificial intelligence has now been identified by the human community as part of everyday life finally opens the door to critical awareness on this subject.
Of course, artificial intelligence concerns industry, work, creation, copyright... and we need to anticipate its future productive uses, in order to stay “up to date”. But to accompany our lives as they integrate this new facet, it seems to me essential to produce a critical thought, i.e. to put ourselves in a position to reflect on what is happening to us, what is changing us, to remain lucid and capable of freedom of thought and action.
What is “critical thinking”? It means questioning, from the outside, practices that have been internalized. To do this, I believe that experimentation, cultural action, play and hijacking are highly effective tools for research, exploration, dissemination and reflection. For me, research is collaborative, and intelligence is collective and creative. This requires good methods of cooperation, between human beings and with machines. Here, I bring together stories of experience, methodological texts and practical ideas. I share concrete ways in which artificial intelligence, like any other tool, can be invested in the service of humanism.
Here are a few openings for critical thinking on AI, in the form of questions: