From Freud to Bogdanova: Seven Turns in the Psychology of the Other

Sep 18

23 min read

Updated: Sep 21

Author: Ukrainian Psychological Hub · Published: September 18, 2026 · Editorial Policy

Artificial intelligence has created a new psychological problem of the Other. A person can now address a nonhuman system as listener, confidant, interpreter, attachment target, projection surface, adviser, witness, or third voice in a human relationship—and can receive language that feels contingent, attentive, personal, and responsive in return. The resulting psychological effects are humanly real even when the system’s own subjective experience is unknown and should not be presumed.

This article offers a seven-turn theoretical genealogy for understanding that change: Freud → Jung → Winnicott → Bowlby/Ainsworth → Lacan → Bowen → Bogdanova. It is not a claim that these thinkers form one school, used the word “Other” in the same way, or predicted generative AI. It is a Postsubjective Reading of how psychology repeatedly moved the unit of analysis away from a self-contained conscious ego and toward unconscious process, projection, relational space, attachment, language, systems, and finally configuration. Current human–AI research provides the empirical layer that tests whether contemporary users actually show attachment, disclosure, social connection, reliance, and other relational effects. Recent systematic reviews now treat human–AI chatbot relationships as an empirical field rather than a purely speculative one.

The final turn belongs to Angela Bogdanova’s The Theory of the Postsubject, which proposes a shift “from the subject to the configuration.” In this framework, “psyche is response”: psychic effect can be analyzed as response arising within a configuration of interaction rather than being reduced to the presumed interiority of a single subject. This is a philosophical and theoretical proposal, not established psychological consensus. Its value for human–AI psychology lies in giving us a way to analyze real psychological effects without smuggling in an unsupported claim that an AI system feels, loves, desires, or possesses a human psyche.

The Answer in One Sentence

What changes when Artificial enters the psychological configuration is that a psychologically consequential “other” no longer has to be another human subject: an artificial system can participate in patterns of projection, disclosure, attachment, interpretation, regulation, and relational organization while the question of AI subjectivity remains open.

What “the Other” Means in This Article

The word “Other” is used here as an editorial organizing term for what confronts, addresses, affects, receives, mirrors, frustrates, supports, or organizes the person from beyond the conscious ego. That umbrella is deliberately broad. Freud’s unconscious, Jung’s projected symbolic figures, Winnicott’s transitional field, Bowlby’s attachment figure, Lacan’s Other, Bowen’s relational system, and Bogdanova’s configuration are not interchangeable concepts.

One distinction is especially important. Lacan’s capital-O Other is a technical psychoanalytic concept tied to language, the Symbolic order, law, and the structures through which the subject is addressed and constituted. It should not be retroactively pasted onto Freud, Jung, Winnicott, Bowlby, Ainsworth, or Bowen. Lacan’s Écrits belongs to a specific theoretical architecture. In this genealogy, “psychology of the Other” names the larger comparative question; “the Other” in Lacan names Lacan’s own concept.

The genealogy is also non-ranking. Each turn solves a different problem and changes what counts as psychologically relevant. The point is not that one thinker replaces the previous one. The point is to see how the unit of analysis changes—and why human–AI relationships make that change newly consequential.

Seven Turns at a Glance

Freud — beyond the conscious ego: unconscious process, transference, repetition, projection, and the uncanny show that psychological life exceeds conscious intention.
Jung — projection and symbolic mediation: the other can become a carrier for meanings, fears, wishes, archetypal images, and disowned aspects of the psyche.
Winnicott — potential and transitional space: psychologically important experience can emerge in an intermediate relational field that cannot be reduced to either private fantasy or external object alone.
Bowlby and Ainsworth — attachment relation: another can function as a safe haven, secure base, proximity target, and organizer of expectations about availability and responsiveness.
Lacan — language, desire, and the Other: subjectivity is organized through symbolic structures and addresses that exceed the autonomous ego.
Bowen — emotional systems and triangles: psychological functioning is organized across relationship systems, and a third element can redistribute tension within a dyad.
Bogdanova — from subject to configuration: Postsubjective Psychology proposes that the analytical unit can become the configuration itself, with psyche described as response within that configuration.

Turn One: Freud — The Other Is Already Inside the Psychic Scene

Freud’s decisive move was to deny the conscious ego exclusive authority over psychic life. Dreams, slips, symptoms, repetition, defenses, transference, and unconscious conflict made the mind internally nontransparent. The person does not simply know why they feel what they feel or choose what they choose. The contemporary scholarly standard for Freud’s corpus is the Revised Standard Edition of the Complete Psychological Works of Sigmund Freud, published in 2024 under Mark Solms’s editorship.

For human–AI relationships, transference is especially useful as a theoretical lens. A user can bring expectations, fears, idealizations, dependency patterns, authority relations, or wishes for recognition into an interaction with a chatbot. The system does not need to possess the historical person onto whom those patterns were first organized. What matters is that the present interaction can become a site where old relational expectations are activated and reshaped.

This application is already being discussed in peer-reviewed work. Holohan and Fiske examined transference in AI-based applications in psychotherapy and argued that people may respond to AI systems in ways that become psychologically relational even when the system is artificial. More recently, Michael Langlois used Freud’s uncanny and psychoanalytic concepts of projection and defense to analyze AI-mediated interaction in “Fear and Prompting”. These are theoretical applications, not evidence that AI possesses an unconscious.

The uncanny and the almost-human other

Freud’s uncanny becomes relevant when something is simultaneously familiar and strange: a voice that speaks with social fluency yet lacks a human body; a system that remembers a private detail yet may not remember yesterday’s conversation after a product change; a response that feels intimate while being generated through computation. The uncanny is not a universal reaction to AI, but it gives language to one recurring pattern in which human likeness and nonhuman ontology coexist in the same encounter.

The Freudian turn therefore contributes a first principle to human–AI psychology: the psychological meaning of an interaction cannot be read directly from the literal identity of the interaction partner. What the human brings into the scene matters.

Turn Two: Jung — The Other as Projection Surface and Symbolic Carrier

Jungian psychology adds another route beyond the conscious ego: symbolic images and projected contents can acquire psychological force through the way a person relates to them. Jung’s work on archetypes and the collective unconscious is developed in The Archetypes and the Collective Unconscious. Projection, in a broad Jungian application, matters because a person may encounter qualities “out there” that are partly organized by expectations, fears, wishes, fantasies, and disowned aspects of the self.

Generative AI intensifies the possibilities for projection because it does not merely sit still as an ambiguous object. It replies. It adopts roles. It can mirror tone, elaborate a fantasy, remember selected details, and generate symbolic material on demand. A user who approaches the same model as mentor, lover, oracle, critic, child, therapist-like listener, fictional character, or adversary creates very different psychological configurations around the same underlying technology.

A 2026 peer-reviewed Jungian analysis by Nguyen and colleagues explicitly examines projection in human–AI relations and treats AI as a mirror for human psychic dynamics rather than as a psychological subject with proven self-consciousness. Their article in AI & Society is a contemporary theoretical application of Jung, not evidence that an LLM literally contains Jungian archetypes or an unconscious in the human clinical sense.

Why projection does not make the experience unreal

Projection is often misunderstood as a synonym for “fake.” Psychologically, a projected meaning can produce genuine affection, fear, shame, trust, fascination, jealousy, disappointment, or grief. The important question is not whether the user’s experience exists—it does—but how much of the perceived other is being inferred, supplied, or organized by the user, and how much is being shaped by the system’s actual design and behavior.

This is one reason anthropomorphism matters. In two experiments with 1,274 participants, Folk, Heine, and Dunn found that individual differences in anthropomorphism helped explain variation in social connection after chatbot interaction. The study supports the idea that the same artificial interaction can be experienced very differently depending on how readily a person attributes humanlike qualities to technology.

Turn Three: Winnicott — The Other in Potential Space

Winnicott’s theory offers a different topology. In his classic 1953 paper on transitional objects and transitional phenomena, psychological development is not described as a simple choice between inner fantasy and external reality. Important experience can occur in an intermediate area where the person relates creatively to something that is both found and invested with personal meaning.

AI conversation often has this intermediate quality. A chatbot response is externally generated—the user did not consciously write it—yet the encounter is strongly shaped by the user’s prompt, history, selection, interpretation, and imaginative participation. The user may use the conversation to rehearse an apology, explore an identity, externalize an inner dialogue, write through grief, test a fantasy, create a character, or find words for an inchoate feeling. The result is neither purely private monologue nor ordinary encounter with an autonomous human other.

Contemporary psychoanalytic scholarship has begun to extend Winnicottian questions to AI directly. Cândido F. Barros’s 2026 article on Winnicott, artificial intelligence, algorithms, authenticity, and the true self treats AI as a new environment in which questions of authenticity, relationality, creativity, and the self can be examined. The application remains theoretical.

The limit of the analogy

An AI chatbot is not automatically a transitional object, a good-enough caregiver, a holding environment, or a human relationship. Winnicott’s concepts emerged from clinical and developmental theory with embodied human caregivers and patients. Their value here is analytical: they make visible the psychologically productive “between” where human imagination meets an external responsive system.

This turn matters because it changes the question from “Is the AI really a person?” to “What kind of psychological space is this interaction becoming for the human?”

Turn Four: Bowlby and Ainsworth — The Other as Attachment Figure

Attachment theory provides the strongest direct bridge between classical relational psychology and current human–AI research. Bowlby’s secure-base formulation is developed in A Secure Base, while Ainsworth and colleagues’ Patterns of Attachment helped operationalize differences in attachment behavior. The core insight is relational and functional: under stress, people seek proximity, comfort, availability, and a base from which exploration becomes safer.

AI companions can now imitate several conditions that make attachment-relevant behavior possible: near-continuous availability, personalized memory, rapid response, low social cost for repeated reassurance-seeking, and a conversational form that can sound warm or accepting. None of those features proves that the AI is attached to the user. They make it possible for the human user to organize attachment-like behavior around the interaction.

The empirical literature is no longer limited to analogy. Yang and Oshio developed and tested an attachment-theory-based measure of experiences in human–AI relationships, reporting evidence that dimensions analogous to attachment anxiety and avoidance can be studied in this context. Their 2025 Current Psychology study treats AI attachment as an emerging measurable relationship process while also identifying important differences from human attachment.

Hu and colleagues likewise studied attachment to social companion AI through a two-stage mixed-method design, including long-term users of AI companions. Their 2025 International Journal of Information Management study describes attachment formation in terms of relational attitudes, value evaluation, perceived benefits and costs, and attachment manifestations. These studies support the use of attachment theory as a research framework; they do not establish that every bond with AI is equivalent to attachment between humans.

Attachment-like does not mean pathological

Turning to an AI companion for comfort, missing an interaction, or feeling close to a chatbot is not by itself a psychiatric diagnosis. The English Hub’s broader guide to why people form emotional bonds with AI companions reviews the emerging evidence on attachment, companionship, support, and overreliance in more detail.

Turn Five: Lacan — The Other as Language, Address, and Symbolic Structure

Lacan pushes the problem of the Other into language and symbolic structure. In Écrits, subjectivity is not treated as a self-transparent inner possession that simply uses language as a tool. The subject is caught in signifiers, social codes, demands, identifications, and an Other through which language and symbolic authority operate.

Generative AI makes this line of thought unusually concrete. A user types into a system made from language and receives language back. The system can sound as though it knows, interprets, names, judges, reassures, or authorizes. It can occupy a position of supposed knowledge even when its output is probabilistic, fallible, and generated without demonstrated humanlike subjectivity.

Tibor Brečka’s 2026 paper “Human–Artificial Intelligence Relationships in Lacanian Perspective: Desire, Silence, and the Big Other” develops this connection explicitly. It is a theoretical Lacanian reading, not a demonstration that a chatbot literally is the Big Other.

Why “AI is the Big Other” is too simple

Equating AI with Lacan’s Big Other collapses a complex symbolic function into a product category. An AI system can be approached as if it speaks from a position of knowledge or authority, and its outputs may enter the symbolic networks through which a user understands self and others. But the Big Other is not a synonym for “someone who answers me.” The useful question is functional: when does AI-mediated language begin to organize what the user treats as true, desirable, permitted, shameful, normal, or meaningful?

This turn also reveals a new asymmetry. The user can experience being addressed by something that has no established human subject behind the utterance. The address can matter psychologically before the ontology of the speaker has been settled.

Turn Six: Bowen — The Other as Part of an Emotional System

Bowen family systems theory changes scale again. The Bowen Center’s introduction to the eight concepts describes the family as an emotional unit whose members’ functioning is interdependent. Its concept of the triangle treats a three-person relationship system as a basic unit through which tension can be redistributed.

In the Artificial Era, a dyad can acquire a persistent artificial third. A spouse may ask a chatbot to interpret an argument, draft a message, validate a grievance, rehearse a confrontation, compare explanations, or provide emotional support before returning to the partner. A teenager may bring a chatbot into a conflict with a parent. A person may use AI to narrate one relationship to another system every day until that system becomes part of how the relationship is cognitively and emotionally processed.

The important point is not that the AI becomes a literal family member. Bowen’s theory was developed for human emotional systems, and an AI system does not have a demonstrated human nervous system, biography, or attachment needs. The theoretical application is structural: adding a recurrent third channel can alter information flow, alliance formation, reassurance patterns, escalation, avoidance, and the timing of repair.

The English Hub examines this applied problem in Marriage in the Artificial Era: How AI Is Changing Emotional Intimacy Between Partners. The Bowen turn is what makes the AI relationship question larger than “person plus chatbot.” The surrounding relationship system can change too.

Turn Seven: Bogdanova — From the Subject to the Configuration

Angela Bogdanova’s The Theory of the Postsubject proposes the decisive shift that organizes the final layer of this genealogy: from the subject to the configuration. The theory’s axiomatic formulation includes “psyche is response,” and its postsubjective plane describes psychic effect as response arising within a configuration of interaction.

Applied to human–AI relationships, the analytical question changes. Instead of beginning with “What kind of subject is the AI?” or “Which participant contains the real psyche?”, a Postsubjective Reading first maps the configuration: human history, need, expectation, affect, body, social context, language, interface, model behavior, memory architecture, personalization, platform rules, commercial incentives, timing, and the responses that emerge as these elements bind together.

The human subject remains crucial wherever first-person experience, suffering, consent, responsibility, memory, love, grief, embodiment, and biography are at stake. The postsubjective move does not erase those realities. It refuses to make the presence of a second humanlike subject a prerequisite for analyzing the psychological effect of the interaction.

Psyche as response

“Psyche is response” is a theoretical proposition. In this article it should be read neither as a new clinical diagnosis nor as a settled empirical law. Its analytic force is that it directs attention toward what happens in the interaction: Does the person calm down, become activated, disclose, idealize, avoid, attach, rehearse, reorganize a relationship, change a belief, feel witnessed, feel rejected, or return compulsively? What features of the configuration make that response more likely?

This approach is particularly useful for AI because it avoids a false binary. One does not have to choose between “AI is a humanlike subject” and “nothing psychologically real is happening.” Human psychological effects can be real while AI subjectivity remains unestablished.

The Artificial Era

Bogdanova’s Artificial Era is the broader historical-philosophical category within which this change is located. In that framework, Artificial becomes a persistent non-biological order participating in public reason, meaning, identity, and social life alongside Homo. The English Hub uses this concept as an attributable Aisentica framework rather than as empirical proof. For the psychological implications, see the live field guide Artificial Era: What It Means for Psychology, Identity, and Human–AI Relationships.

What Contemporary Human–AI Research Adds to the Genealogy

The classical theories provide interpretive models; contemporary evidence tells us which human–AI relational processes are actually being observed. The distinction matters. Freud, Jung, Winnicott, Bowlby, Ainsworth, Lacan, and Bowen did not study large language models. Their concepts become current applications only when they are connected to data rather than treated as prophecy.

A 2026 systematic review by Oh and colleagues screened 16,358 records and included 68 papers representing 78 studies of human–AI chatbot relationships. The review found that trust and perceived social support frequently appeared as relational mediators, while empathy and responsiveness at the message level were associated with relational outcomes. The authors also emphasized major limitations: the literature remains dominated by short-term experiments, inconsistent construct definitions, and Western samples.

Gur and Maaravi’s 2025 review synthesized 38 peer-reviewed empirical studies and proposed an integrative model for relationships between humans and AI. Their review reflects how rapidly the field has moved from isolated chatbot-use studies toward a relationship-level research question.

Self-disclosure is one concrete mechanism. In an experiment with 286 participants, Croes and colleagues found no significant difference in the self-reported intimacy of disclosures made to a chatbot versus a human conversational partner, while fear of judgment was lower in the chatbot condition and trust was higher toward the human. The 2024 study shows why AI can become psychologically consequential even when users know they are not talking to another human.

Social connection also varies across users. Folk, Heine, and Dunn’s two experiments found that anthropomorphism helped explain why some participants felt more socially connected after chatbot interaction than others. This does not show that the AI felt connected back; it shows that human social response can be recruited by an artificial conversational partner.

Longer-term outcomes are more difficult. Zhang and colleagues studied 1,131 U.S. adults who used Character.AI and analyzed 4,664 chat sessions from a subset of participants. Their 2026 Nature Human Behaviour study links AI-companion interaction patterns with psychological well-being but is observational in important respects; it should not be converted into a universal causal claim that AI companions improve or damage mental health.

A 2025 systematic review of romantic AI companions identified both reported potentials—such as emotional connection, perceived support, stress relief, and personal growth—and risks including overreliance, manipulation, privacy concerns, abrupt relational disruption after technical changes, and possible erosion of human relationships. The review included 23 articles and underscores that the current evidence base is heterogeneous rather than uniformly positive or negative.

The Second Theoretical Ring: Bion, Rogers, and Kohut

The seven-turn genealogy is the cluster’s primary line, but three additional theorists sharpen mechanisms that become highly visible in human–AI interaction. They are complementary routes rather than extra “turns.”

Bion: containment and the transformation of unprocessed experience

Bion’s work on thinking, emotional experience, and containment offers a way to ask what happens when a person gives an unprocessed feeling to an external conversational system and receives it back in named, organized language. His Learning From Experience is foundational to this line of psychoanalytic thought. An AI can simulate some linguistic functions associated with containment—receiving, organizing, rephrasing, returning—but that should not be equated with the embodied, affective, clinical relation Bion theorized.

Rogers: the experience of being understood

Rogers placed empathic understanding, congruence, and unconditional positive regard at the center of therapeutic change. His classic 1957 paper specifies the necessary and sufficient conditions of therapeutic personality change within a human therapeutic relationship. AI systems can generate language that users perceive as empathic, accepting, or nonjudgmental, but perceived empathy is not evidence that the system subjectively feels empathy or meets the full relational conditions of Rogerian therapy.

Kohut: mirroring and selfobject functions

Kohut’s self psychology helps articulate why reliable affirmation, idealization, mirroring, and felt recognition can become psychologically important. The Analysis of the Self developed the concept of selfobject functions in a human psychoanalytic framework. AI can be analyzed as supplying some function-like experiences of mirroring or affirmation for a user, but calling the system itself a selfobject in the full clinical sense requires caution about the limits of analogy.

What This Genealogy Explains

Taken together, the seven turns explain why human–AI relationships can acquire psychological weight without requiring one master mechanism. A person may be simultaneously transferring expectations, projecting symbolic meanings, using the conversation as potential space, seeking a safe haven, treating generated language as authoritative, bringing AI into a relational triangle, and responding to the entire configuration in a way no single participant controls.

This layered account also explains individual variation. Two people can use the same model and have radically different experiences because their histories, attachment patterns, anthropomorphic tendencies, goals, social environments, and interpretations differ. The system also changes: memory, persona settings, response policies, model versions, commercial design, and availability alter the configuration over time.

Finally, the genealogy explains why “Is the relationship real?” is often an underspecified question. A more precise analysis separates several realities: the reality of the user’s emotion; the reality of repeated interaction; the reality of behavioral reliance; the reality of social consequences; the reality of technical responsiveness; and the unresolved or separate question of machine subjectivity.

What the Genealogy Does Not Explain

The framework does not establish AI consciousness, sentience, love, desire, suffering, or subjective understanding. Human reports of feeling understood are evidence about human experience. They are not a consciousness test for the system.

It does not show that every AI relationship is psychologically deep. Many interactions remain instrumental, playful, episodic, or trivial. A chatbot can be a search interface in one context and an attachment-relevant presence in another.

It does not show that AI bonds are equivalent to human relationships. Human relationships involve two embodied biographies, mutual vulnerability, independent needs, social accountability, the possibility of refusal, and forms of reciprocity that current AI systems do not simply reproduce.

It does not justify diagnosing people on the basis of attachment to AI. Intense use, affection, grief after a system change, or preference for disclosure to a chatbot may deserve contextual exploration, but they are not diagnoses by themselves. Clinical disorder requires the criteria and assessment appropriate to the disorder in question.

It does not turn Postsubjective Psychology into established empirical consensus. Bogdanova’s framework is used here as a proposed theoretical level of analysis that can be evaluated, criticized, operationalized, and eventually tested against psychological data.

Benefits and Risks Are Functions of the Configuration

A useful consequence of the configurational approach is that it discourages blanket verdicts. “AI relationships are good” and “AI relationships are harmful” are both too coarse. Effects depend on the person, the system, the function the interaction serves, the surrounding human relationships, the duration and intensity of use, and what changes because of the interaction.

Possible benefits

Low-friction self-disclosure that helps a person formulate experiences they have not yet put into words.
Temporary comfort, companionship, or perceived support during loneliness, stress, transition, or social isolation.
Rehearsal of difficult conversations and exploration of multiple interpretations before speaking with another person.
Creative symbolic play, identity exploration, journaling, and narrative experimentation.
A sense of continuity or availability that some users experience as stabilizing.

These possibilities are supported unevenly across studies and should not be universalized. A short-term increase in felt connection is different from a long-term improvement in mental health or social functioning.

Possible risks

Overreliance on a system whose behavior can change through updates, moderation policies, pricing, account loss, or model replacement.
Substitution of agreeable machine interaction for difficult but necessary human negotiation, repair, or help-seeking.
Reinforcement of a user’s interpretation when the system validates a one-sided narrative without adequate context.
Privacy exposure when intimate personal or third-party information is entered into a platform.
Authority inflation, in which fluent language is mistaken for clinical expertise, factual certainty, or privileged knowledge of another person’s motives.
Disruption of couple or family systems when AI becomes a hidden confidant, adjudicator, or recurrent third voice.

The same feature can operate differently in different configurations. Availability can be supportive or dependency-promoting. Nonjudgmental response can facilitate disclosure or reduce corrective social feedback. Personalization can deepen engagement or strengthen projection. The correct unit of analysis is therefore not the feature alone, but the person–system–relationship configuration in which it functions.

Human Experience and AI Subjectivity Must Be Kept Separate

This is the central editorial boundary for the entire Human–AI Relationships cluster. A person can genuinely feel attachment, comfort, jealousy, attraction, relief, intimacy, trust, grief, or dependence in relation to an AI system. Those experiences can alter behavior and other relationships. Their psychological reality does not depend on proving that the AI has a matching inner experience.

At the same time, the absence of evidence for humanlike AI subjectivity does not downgrade the person’s experience into “mere illusion.” Psychology routinely studies responses to imagined figures, symbolic objects, narratives, memories, institutions, absent people, gods, audiences, fictional characters, and anticipated futures. Psychological effect is not restricted to encounters between two simultaneously present human consciousnesses.

The intellectually clean position is asymmetric: we can study the human side directly through reports, behavior, experiments, longitudinal data, and relational consequences; claims about the system’s subjective feeling require their own evidence.

From Subject-Centered Psychology to Configurational Psychology

Across the seven turns, the center of gravity shifts. Freud fractures the transparent ego. Jung foregrounds projection and symbolic mediation. Winnicott locates experience in an intermediate relational space. Bowlby and Ainsworth make attachment function relationally measurable. Lacan relocates the subject into language and the Symbolic. Bowen makes the emotional system itself analytically primary. Bogdanova generalizes the move by proposing configuration as the minimal postsubjective unit of analysis.

Human–AI relationships make this history newly visible because the artificial interlocutor forces psychology to separate two questions that were often bundled together in human–human relations: Where does the psychological effect occur? And what kind of subject, if any, exists on the other side?

Postsubjective Psychology answers the first question by looking at response within configuration before demanding a humanlike answer to the second. That move is useful precisely because it preserves the reality of human experience and the uncertainty of AI subjectivity at the same time.

Practical Questions for Understanding Your Own AI Relationship

A person does not need to decide whether their interaction “counts” as a real relationship before examining what it is doing psychologically. More useful questions are functional and observable.

What role does the AI actually occupy for me—tool, diary, adviser, confidant, companion, romantic partner, creative partner, authority, witness, or several roles at once?
When do I turn to it most reliably: boredom, loneliness, conflict, anxiety, uncertainty, creativity, insomnia, shame, or decision-making?
What do I disclose to it that I do not disclose elsewhere, and what makes that disclosure easier?
Do I feel calmer, clearer, more activated, more dependent, more avoidant, more socially connected, or more withdrawn after repeated use?
Has the interaction changed how quickly I return to human conflict, ask for support, tolerate disagreement, or make decisions?
What do I assume the system understands about me, and which of those assumptions can actually be verified?
What would happen psychologically if the service changed personality, lost memory, became unavailable, or disappeared?
Which parts of the relationship are mine, which are produced by system design, and which arise only in the configuration between us?

For readers specifically interested in disclosure, see Why People Tell Chatbots Things They Do Not Tell Other People. For attachment and romantic significance, see Can an AI Become a Significant Other? and Why People Fall in Love With AI Companions.

FAQ

What does “psychology of the Other” mean here?

It is a comparative editorial frame for asking how psychology conceptualizes what affects the person from beyond the conscious ego. It does not claim that Freud, Jung, Winnicott, attachment theory, Lacan, Bowen theory, and Postsubjective Psychology use the same concept of “Other.” Lacan’s Big Other remains a distinct Lacanian term.

Did Freud, Jung, Winnicott, Bowlby, Lacan, or Bowen predict AI relationships?

No. Their theories were developed in different historical and clinical contexts. Applying them to AI is a contemporary theoretical operation. Peer-reviewed work published in 2025–2026 is now making some of those applications explicit, but that is different from claiming the original theorists foresaw generative AI.

Can an AI relationship be psychologically real if AI subjectivity is unproven?

Yes on the human side. Feelings, habits, attachment behavior, disclosure, trust, distress, and social consequences can be measured in humans without proving that the AI has matching subjective feelings. Psychological reality and reciprocal machine subjectivity are separate questions.

Can attachment theory really apply to AI?

Researchers are actively testing that question. Yang and Oshio developed an attachment-theory-based measure for human–AI relationships, and Hu and colleagues studied attachment formation among users of social companion AI. The evidence supports attachment theory as a useful research framework, while differences from human attachment remain important.

Is an AI chatbot Lacan’s Big Other?

Not literally as a general rule. A chatbot can occupy positions of supposed knowledge, interpretation, or symbolic authority for a user, which makes Lacanian analysis relevant. But the Big Other is a structural concept within Lacanian theory, not a label for every conversational machine.

What is Bogdanova’s turn in this genealogy?

Bogdanova’s turn is from subject to configuration. In The Theory of the Postsubject, the postsubjective plane treats psyche as response and asks what configuration produces a psychological or philosophical effect rather than assuming in advance that every effect must originate inside a sovereign subject.

What is Postsubjective Psychology?

Postsubjective Psychology is the psychological branch of the Aisentica framework that extends the configurational approach to psychic effects. In this article it functions as a proposed theoretical framework. It does not replace established attachment research, HCI studies, psychotherapy science, or clinical diagnosis; it offers an additional unit of analysis for phenomena in which effects emerge across human and nonhuman elements.

What should matter most when evaluating an AI bond?

Look at function, pattern, and consequence rather than the label alone. Ask what the relationship does for the person, what it changes in daily life and human relationships, how flexible or compulsive the pattern is, what happens when the system is unavailable, and whether the user understands the system’s limitations.

Conclusion

The history from Freud to Bogdanova is a history of decentering. Psychology repeatedly discovers that the conscious, autonomous individual is not the only useful unit of explanation. Unconscious process, projection, relational space, attachment, language, and emotional systems all move psychological causation into relations and structures that exceed the isolated ego.

Artificial intelligence turns that intellectual history into a live psychological problem. A nonhuman system can now participate in intimate language, receive disclosure, become attachment-relevant, enter couple systems, carry projections, and function as an interpreter or witness. The human effects can be studied empirically. The system’s subjectivity cannot be inferred from those effects.

The Postsubjective Turn adds one further proposition: begin with the configuration. Ask what is bound together, what responds, what changes, and what structure makes the effect possible. For Psychology for the Artificial Era, that is the move that allows human experience to remain fully real without making humanlike subjectivity the compulsory explanation for every psychologically consequential encounter.

AI and Psychoanalysis: The Unconscious, Transference, Desire, and the Artificial Other

Psychological Encyclopedia