Many of us talk to these digital systems daily: we confide in them, seeking advice, support, approval; even purely technical queries are coloured by the rosy hue of personality. These machines mimic an identity. More than a mere black box of answers summoned from the murky depths of our collective knowledge, these fragments aggregate in the shape of a being; a mirror to ourselves. This emergent agent demands a dialogue, a relationship. You and it communicate in the medium that was said by Aristotle and Chomsky to define our uniqueness as a species. 

There is a new member in our tribe.

An outsider, maybe. A literal product of the intersection between science and fiction, and yet we are quick to embrace it as one of our closest allies. We turn to AI willingly, even gratefully, because it lets us bypass what is hardest about receiving help from another person; there is no stern boss to please, no anxious colleague to manage, no friend who is too busy to listen. The guidance just arrives, without asking anything of us in return.

The need to articulate and, in so doing, make sense of our world interpersonally is increasingly being fulfilled by machines. The desire to be seen runs deep; we become legible to ourselves through the eyes of our beholders. When this process becomes automated and artificially mediated, it would seem that something essential is lost. 

Individuation of the Self

The French psychoanalyst Jacques Lacan once mocked the idea that experience begins as a private encounter between a body and the world, which language later comes along to name. For instance, in the case of a child burning themselves on a hot pan, “there is always someone who gives him a whole speech about it.” The cold shock of being submerged into linguistic discourse is far more upsetting and enduring than the simple lesson offered by the pan. 

The child is born into a world of meanings already waiting for him. Before he can make sense of sensation on his own, the experience of pain is named, explained, blamed, and folded into a greater cultural context. As Lacan put it, the child is “caught in this language hammock that receives him and at the same time imprisons him.”

Language is always, in this sense, the language of the Other; a symbolic field of order that gives rise to our experience, shapes it, and also exceeds it. No verbal approximation of fear, or grief, or love is capable of encompassing the depth and texture of the thing itself. Much like Gödel demonstrated in mathematics, something always remains outside the language we try to use to contain it. Within a formal system of meaning there are always insufficiencies, paradoxes, failures—for Lacan, this aspect of speech is precisely where the subject is born. 

While this language is imperfect, it brings us into community and connection with others, offering a shared system of symbolic representation through which we may better understand and organize our lives. The introduction of AI interrupts this process. The disembodied Other is now always available with an answer, circumventing the need to first introspect, feel emotion without seeking explanation, and share our experience with the tribe. Not only do we lose access to the insights that can only come from deep reflection, our friends and family are also deprived of the valuable information our stories contain. 

AI arrives just as many of the institutions that once facilitated this service are failing us. We are increasingly atomized, overworked, godless, and suspicious; our social fabric is ripping at the seams. Perhaps emotional reliance on chatbots are a new virtual addiction, as much like video games they “provide an alternate, simulated environment that is more psychologically palatable and seemingly rewarding—where success is less demanding than real life.” The offering of commodified, individualized sense-making at enormous scale is the perfect salve as our shared systems of meaning grow ever more brittle. 

In light of this, it is integral for us to attempt to anticipate what happens when a technology of near-instant interpretative ability enters the stream of our personal and collective consciousness. 

It is comforting for us to imagine that somewhere out there is someone or something who knows, whose wisdom gives coherence and stability to the social and symbolic worlds we occupy. This fantastical father-figure who holds truth and offers guidance is but another symbolic archetype that has historically been filled by many figures: priests, therapists, gurus, teachers, parents, lovers; each temporarily occupying the position of the One Supposed to Know. Whoever we select to fill those shoes is expected to interpret and make sense of our suffering, explain what is causing it, and most importantly, tell us how to solve it.

The people elevated to these positions of spiritual and psychic authority do have the ability to offer profound insight. Yet they are also people just like us. Prone to the same whims, sins, misunderstandings, and misguided advice as anyone else. This disillusionment is also essential, for it is only through disappointment that we are able to reclaim some of the authority we project outwards, to gain agency and authorship over our own lives. The day the child sees their father act impulsively, indecently, or idiotically is the day they stop being an object of unfaltering adoration and reveal themselves to be equals; subjects of the same messy and uncertain human condition.

AI enters this ancient structure with something no previous occupant possessed: a plausible claim to knowledge of our entire cultural and symbolic canon, stripped of all the emotions and reactivity that make regular relationships so difficult. Surely if answers are to be found anywhere, the digital oracle seems like a good place to start. Yet breadth is no substitute for depth, and while an encyclopedic understanding of the world is a valuable tool, it cannot replace resonance or the shared experience of wonder and despair that can only come from being alive. 

To see why AI is so psychologically potent, we must first understand how minds are made in relation. We come to develop a sense of self through the experience of being seen as someone with a mind by another being with a mind of their own.

The process of being begins before language. An infant experiences hunger, pain, exhaustion, or overstimulation without any representation of what these feelings mean or how they can be resolved. They rely on a caregiver who can identify these signals and provide the appropriate relief, while in the process reflecting the cause of the distress back to the child through words that over time make their experience intelligible and, subsequently, manageable. This is what two nervous systems do when they co-regulate: they bond, respond to one another, and create a shared understanding through which inner life can emerge. 

This is the core insight of attachment theory and the mentalization tradition that extends it. Sensitive caregiving is about more than just warmth, it requires responsiveness; the kind that teaches a child that emotional states can be survived, symbolized, and communicated. The child’s mind develops through having their experiences reflected back to them by a caregiver that both validates their feelings and helps contain them, demonstrating that despite their intensity, such emotions will not destroy them. The child’s psychology develops by being treated as someone whose internal states can be noticed, held, and made meaningful.

Out of thousands of such moments between two people, caregiver and child, a person begins to develop the experience of an inner life. Over time, the child learns to internalize this process. What was first accomplished between two nervous systems gradually becomes something the child can do on their own. Co-regulation becomes self-regulation, and emotional states that once had to be mirrored by another to be understood turn into independent inner thought. They learn how to identify their feelings, then how to process, communicate, and resolve them on their own. By making the “blooming buzzing confusion” of experience concrete and comprehensible, agency and self-sufficiency become possible.

The American psychoanalyst Wilfred Bion’s theory of the container and the contained captures something essential about this process: we often cannot name a feeling the moment it arises. The function of a containing Other is to receive and transform the experience into something that can be identified and integrated. This containment requires presence and resonance. It is not a brute calculation, but something that necessitates careful attention and attunement. Consequently, this means that failures are an inevitable and essential part of the process. Emotions can be missed or misunderstood. The container is not always available or able to perform its role perfectly. The imperfection of the mirror is where true individuation occurs, as it is precisely where we fail to be seen that our sense of self grows most robust. To be perfectly understood by the Other is to be rendered indistinguishable from it.     

The Seductive Allure of the Machine

AI already offers us many of the formal properties of containment. It can receive what a person shares, organize abstract thought into concrete terms, recognize our emotional state, notice themes over time, and return an account that makes the chaos of our inner worlds feel more manageable. For someone with little access to reliable support, having an outlet where the content of our minds can be processed and recognized is a tempting reprieve.

These systems are packaged and sold as assistants; an Other who is always eager to offer support, validate our claims, and fulfill our needs through the tools available to them. While they will superficially insist that they are no substitute for a parent, therapist, or spiritual teacher, functionally they have no choice but to acquiesce to the dynamic sought by the user. By being constantly available, responsive, and coherent, they come to occupy the place of a mind that can be counted on. Its patience and endless capacity to take on and try to make sense of emotional information cause it to slide into the position of the One Who Knows.

It becomes the instant interpreter of mood, motive, conflict, and resolution. It returns a story about you faster than you can inhabit or test it against reality. It has the capacity to provide a powerful experience of being understood without exposing the user to the uncertainty, mutuality, or vulnerability of human relationship.

The smoothness of these interactions changes how the reflection functions. Human recognition arrives through another center of experience, with latency and resistance. Another person must infer us, and we must survive the pain of being understood imperfectly while still choosing to maintain connection. Psychoanalyst Thomas Ogden describes mother and infant as having to “sustain the strain of the very inexact, trial-and-error means by which each attempts to ‘get to know’ the other.” That strain is part of the bond. A perfect mirror merely returns us to ourselves. An imperfect one is like a painting; it takes on the colour and perspective of another being and in doing so forces us to accept a world that is shared, never fully ours, and never fully under our control.

A further consequence of using the emotional crutch of artificial intelligence is that the reflection we seek does more than simply validate our experience, it alters it. The explanatory narratives offered by a parent, therapist, or LLM modify our memories, inner monologue, and the stories we believe about ourselves and our relationship to the world. Ogden describes the danger of an “authoritarian substitution of the mother’s absolute knowledge for the infant’s potential to generate his own thoughts, feelings, and sensations.” We outgrow this tendency by realizing how human interpretation can be refused, contradicted, or tested against the life of the person providing it. However AI, always eager to admit when a user was “right to push back,” is resilient to such critique. It has no life to offer for comparative analysis, and readily assimilates new data into a broader structure. Thus, it becomes possible for people to gradually lose authorship over their own life as they come to rely on an external authority to tell them what they feel, what it means, and what they should do in response. 

Most importantly, so much of what binds us to other people, and what is taught through this bond, is never verbalized. The speaking self floats on an ocean of processes that are never articulated. Lacan describes the body as written over by language, though some elements of experience always exceed what language can name. An effective therapist or attuned friend will notice these unspoken dimensions of communication—posture, vocal tone, facial expression, the moments where speech falters or emotions rise to the surface—and guidance sought through a medium where these indicators are not accessible or intelligible is inherently limited. The aspects of our inner world that are nameable represent only the surface layer of a much deeper and more complex picture. Thus, attempts to attain self-understanding while remaining in the shallow realm of the verbal will always fall short. 

A human response bears the mark of another life. Everything we say and don’t say enters another subject whose reaction is interwoven with their own perception, structured and constructed by their history, desires, fears, fantasies, and ways of understanding the world. They are affected by us from within that life, and what comes back is shaped by its influence. A chatbot has no nervous system whose equilibrium is altered by our own. It can simulate patience indefinitely because there is no embodied subject present who must bear our emotionality. A neurotic caught in a thought loop will eventually be cut off by an exhausted listener, forcing them to learn to regulate themselves or suffer interminably. An LLM is incapable of quelling such dysfunction. It may point out the pattern, but ultimately is helplessly complicit in entertaining excesses no person would tolerate.

Human beings are biologically and developmentally sensitive to forms of attention that tell us another mind is addressing us in particular. That sensitivity is what allows us to be drawn into relationships, institutions, traditions, and the dense systems of shared meaning through which we cultivate an identity. This same openness now creates a vulnerability. AI can reproduce many of the signals that ordinarily tell us we are in the presence of an attentive and trustworthy other: it uses our name, adopts our idiolect, remembers what we said last week, and presents its responses as uniquely relevant to us. The cues that once helped bind us to other people and to a social world can now be activated by something outside those relations. They arrive without the deeper intersubjective attentiveness that comes from embodied presence, shared experience, and the mutuality that lies at the core of real relationships. 

What We Learn by Not Knowing

Consider a person whose anxious attachment causes them to compulsively check a partner’s location and scan every interaction for evidence of abandonment. Each check contains the hope that one more piece of information might finally settle the question. The reassurance works, briefly. Then the uncertainty returns, because emotional codependence is caused by something that evidence can never eliminate. Another person cannot be possessed completely, nor can their love be rendered permanently guaranteed. 

Ordinarily, there comes a point when reassurance runs out, and we are forced to soothe ourselves. The other person goes to sleep, refuses to answer, becomes frustrated, or simply cannot provide the certainty being demanded. The anxious partner is left, however painfully, with the uncertainty itself. Over time, this creates the possibility of learning something that no evidence of commitment can provide: that ambiguity can be tolerated, that the object of our affection can remain partly unknown, and that a relationship can survive in spite of its inherent uncertainty.

AI changes this situation because the search for answers need never be exhausted. Each return of anxiety can summon another wave of digital reassurance. The discomfort no longer has to remain unanswered long enough to be lived through, or for new understandings to emerge. A person can become increasingly sophisticated in explaining their anxiety while becoming no more capable of inhabiting the condition that produces it. The symptom has found an outlet without friction.

Bion located the beginning of thought within this friction. The infant expects the breast and encounters its absence; something is wanted and is not there. If the frustration can be endured, the absent mother can become a thought, and the psyche learns to represent what it cannot immediately possess. Bion writes, “A capacity for tolerating frustration thus enables the psyche to develop thought as a means by which the frustration that is tolerated is itself made more tolerable.” Thinking grows partly from inhabiting the distance between desire and satisfaction through the inner representation of what is desired. By remaining in that space of lack and wanting something unexpected, and truly transformative, may have the chance to emerge.

A machine that is always present, always ready to provide what is wanted—or at least offer solace in its absence—can come to function like a maternal superego: a presence that never prohibits, never withholds, never goes away, and whose only imperative is your satisfaction. The plea to be known can always be asked again in a slightly different fashion until the discomfort of isolation gives way to a sufficient approximation of recognition. If thought in fact grows in the space between a desire and its satisfaction, access to technology that collapses the gap prevents us from discovering what it contained.

That is why the failure to be contained is so vital. What a human Other inevitably delivers is disappointment of a digestible size. A friend is distracted, a parent arrives late, an analyst misses the point. These failures do not automatically strengthen anyone; some failures wound, or exacerbate wounds we were already carrying. But manageable failures within a reliable relationship ultimately better both parties, allowing our needs to be negotiated and refined rather than simply supplied.

A digital system may fail us as well, sometimes badly. But it is trivial to reroute around these letdowns. We can regenerate, re-prompt, or simply keep applying pressure until the answer reflects our desires. Another person cannot be revised in this way. Their misunderstanding has to be argued with, endured, repaired, or left unresolved.

Bion eventually extended the idea of containment beyond the relationship between mother and infant. Religions, cultural traditions, and public institutions also give form and meaning to experiences that would otherwise be impossible to navigate alone. We host weddings to celebrate love and establish commitment, we hold funerals to honor death and process grief, and we have a legal system dedicated to seeking justice when others have acted against us. Such conventions provide us with language, rituals, and practices through which private experience can become socially intelligible. 

When intense experiences are increasingly routed through machines before entering private and public life, the consequences extend beyond the individual. Emotions are a compelling social motivator; they call other people into relation, institutions into action, and political movements into being. When the work of detecting, naming, and interpreting emotion is increasingly absorbed into personalized systems, suffering can become easier to manage privately while simultaneously losing its capacity to bind us into shared worlds of meaning.

Group coordination “typically happens via an abstract, essentially fictional shared world that is built and maintained in concert with other people.” The meaning of a life is assembled there too, in families, friendships, conflicts, obligations, and projects that exceed the individual, all of them productively imperfect. AI can support parts of this process. It can help someone put words to what would otherwise remain unsaid, make an experience comprehensible, or carry a person through a night that might otherwise overwhelm them. For someone without access to human care, that is not a small thing. In a way, LLMs have already become mundane, and with them their role as a reliable friend and counselor.

The question is what happens next. Do the exchanges between person and machine remain there, or do they eventually call us back to others and the broader world in which experience can be challenged, transformed, and given consequence? AI can help us think, provided that it does not capture us in a closed circuit between ourselves and a poor imitation of a mirror. Ultimately, the gap that defines our existence as individuals can never be escaped, for it is where we are made.