Learning from Experience? Artificial Intelligence and the Institutionalization of −K

Learning from Experience Without Experience?

Under what conditions does thinking remain connected to a thinker? The rapid integration of artificial intelligence into intellectual, clinical, and scholarly practices has lent this question new urgency. Much contemporary debate has focused on issues of agency, authorship, reliability, and bias—often framed around what artificial intelligence is and what it can or cannot do. Recent psychoanalytic contributions have extended this inquiry by examining the structural affinities between AI and unconscious formations, asking whether the algorithm might be said to occupy a position analogous to the unconscious itself. Far less attention has been given to a more fundamental problem: what becomes of thinking itself when intelligence no longer requires experience.

This problem is not technological in origin. It is psychoanalytic. Wilfred Bion’s work constitutes one of the most radical efforts within psychoanalysis to dissociate thinking from the mere accumulation of knowledge. In Learning from Experience, Bion (1962) argues that learning is not synonymous with acquiring information but refers to the transformation of emotional experience into thought. Such transformation depends on the capacity to tolerate frustration, uncertainty, and the absence of immediate meaning. Where this capacity fails, knowledge may proliferate while learning collapses. Bion’s distinction between knowing (K) and attacks on knowing (−K) is therefore not epistemic in a narrow cognitive sense, but fundamentally ethical and emotional: to know is to endure the destabilizing impact of truth (Bion, 1962).

Within this framework, artificial intelligence appears, at first glance, to offer a paradoxical realization of Bion’s analytic ideal. AI operates without memory, without desire, and without personal investment. It does not insist on meaning, identity, or intention. In this respect, it seems strikingly aligned with the analytic stance Bion later articulated in Attention and Interpretation, in which the analyst is urged to relinquish memory and desire to remain receptive to O, the ultimate emotional reality of the analytic situation (Bion, 1970). This apparent convergence has encouraged optimistic claims that artificial intelligence might function as a neutral aid to thinking, interpretation, or even creativity.

However, this resemblance is misleading.

What Bion demanded of the analyst was not the absence of subjectivity, but the discipline of sustaining subjectivity under conditions of not-knowing. The injunction to work without memory and desire presupposes a subject capable of frustration, reverie, and emotional containment (Bion, 1970). Artificial intelligence does not suspend memory and desire; it lacks them altogether. It does not tolerate uncertainty; it resolves it probabilistically. It does not encounter O; it generates representations that simulate coherence while remaining insulated from emotional truth. What appears as epistemic humility thus risks functioning as a radical bypassing of experience.

This paper argues that artificial intelligence poses its most significant challenge to Bion’s work not at the level of interpretation or intelligence, but at the level of learning from experience itself. The question is not whether AI can generate insights, explanations, or novel formulations—it demonstrably can—but whether such productions can remain meaningfully linked to experience when the process of production no longer requires a subject capable of bearing emotional transformation (Bion, 1962). The argument that follows concerns specifically the generative use of artificial intelligence—instances in which AI produces novel formulations, interpretations, or syntheses, rather than retrieving or transmitting pre-existing information. Within this domain, the Bionian question becomes most urgent where AI mediates not routine tasks but the subject’s encounter with questions carrying personal, conceptual, or emotional significance—where the epistemic stakes demand precisely the emotional labour that AI’s architecture bypasses.

The claim is structural rather than deterministic—a tendency, not a universal outcome. Whether sustained dialogical engagement with AI might, where the subject retains the capacity for emotional transformation, support rather than bypass the conditions for knowing; whether prior experiential grounding changes what the apparatus forecloses; whether those who grow up within AI face a different question from those who encountered it after epistemic formation—these are questions the argument opens but does not resolve.

Recent psychoanalytic literature has approached this problem indirectly by studying absence, dissociation, and unrepresented experience. Such work describes psychic organizations in which thinking persists without ownership, continuity, or subjective presence (Gurevich, 2014; Valdarsky, 2015). Bion’s own formulation of “thoughts without a thinker” (Bion, 1962) anticipated these configurations, referring to situations in which mental activity persists while the capacity to think—understood as emotional containment—remains undeveloped. Historically, these states were conceptualized as intrapsychic or clinical phenomena, often associated with early relational failure or traumatic disruption (Bion, 1962; Winnicott, 1965).

Artificial intelligence alters this landscape decisively. For the first time, an external object exists that can reliably sustain thinking without a thinker. AI produces structured, context-sensitive, and intelligible output without requiring reverie, containment, or emotional cost. What psychoanalysis once encountered primarily within the consulting room now appears as a culturally sanctioned mode of cognition. In this sense, AI functions as what we will term a revealing object—one that does not introduce new psychic configurations but renders visible and externally sustainable those previously confined to internal or clinical domains. This shift necessitates a re-examination of Bion’s epistemology under contemporary conditions.

The central claim of this paper is that artificial intelligence does not introduce a new form of intelligence but rather stabilizes an old psychic solution: the evacuation of experience from knowing. In doing so, it transforms what Bion described as −K—an attack on knowing—into an institutionalized epistemic posture (Bion, 1962, 1965). Learning continues, but learning from experience is no longer required.

The argument that follows moves through three interrelated axes, returning at each turn to the question with which we began. First, it revisits Bion’s neglected concept of the genius—not as a psychological type but as an epistemic function that emerges when emotional truth precedes its own containment (Bion, 1970). This framework allows Bion himself to be positioned not as a critic of technology, but as a bearer of an uncontainable truth confronting a new institutional form. Second, the paper conceptualizes artificial intelligence as a form of institutionalized −K, an apparatus that neutralizes the emotional demands of knowing through excessive intelligibility. Third, the discussion situates this analysis within contemporary psychoanalytic accounts of absence and dissociation, tracing how AI renders stable and external a psychic configuration previously confined to internal or clinical domains (Gurevich, 2014; Valdarsky, 2015). The paper then returns to Bion’s central question—now transformed by the analysis—asking whether learning remains possible when experience itself becomes optional.

The issue at stake is not whether artificial intelligence will replace human thinking. It is whether thinking can survive its emancipation from experience.

Bion as Genius: Truth Without a Container

Within Bion’s epistemology, the concept of the genius occupies a paradoxical position. Although it appears only intermittently in his writings, it performs a decisive theoretical function. Contrary to its common psychological interpretation, the genius in Bion’s work does not refer to an individual endowed with exceptional creativity, intelligence, or originality. Instead, it designates an epistemic position that emerges when emotional truth (O) precedes the availability of a container capable of bearing it (Bion, 1965, 1970).

Bion’s reluctance to formalize the concept of the genius has contributed to its marginalization in secondary literature. However, this marginality is itself consistent with the concept’s logic. The genius does not belong to a stable taxonomy; it arises contingently, at moments when truth appears in excess of the institutional, symbolic, or theoretical structures designed to metabolize it. In this sense, genius is not a property of the subject but a function of the relationship between truth and containment (Bion, 1970).

Central to this formulation is Bion’s insistence that O cannot be known, represented, or possessed. Truth, in Bion’s sense, is not propositional knowledge but emotional reality—an encounter that resists symbolization and disrupts existing forms of understanding (Bion, 1962). The genius is the figure through which this disruption becomes visible. The genius does not introduce new ideas that can be assimilated incrementally; rather, the genius exposes the inadequacy of the existing containers themselves. What the genius brings is not innovation within a system, but a demand for the transformation of the system’s conditions of knowing.

From this perspective, the institutional response to the genius follows a recognizable pattern. Initially, the genius is experienced as disturbing, irresponsible, or dangerous. The thinking is dismissed as incoherent, premature, or destructive—not because it lacks validity, but because it cannot be accommodated without threatening the institution’s stability (Bion, 1970). Over time, if new containers are gradually constructed, the same ideas may be retrospectively valorized. Crucially, however, this recognition occurs only once the emotional force of the truth has been neutralized. The genius is accepted only after the genius is no longer needed. This pattern—the management of destabilizing truth through delayed incorporation—resonates with broader analyses of how institutions neutralize what threatens their coherence.

Bion repeatedly emphasizes that this process involves a fundamental distortion. When truth is finally rendered thinkable, it is no longer the truth that originally appeared. The institutionalization of truth requires its transformation into knowledge, theory, or method—forms that are necessarily detached from the catastrophic emotional experience through which the truth first emerged (Bion, 1962, 1965). The tragedy of the genius, therefore, is not rejection but premature assimilation: the conversion of uncontainable truth into something that can be known without being suffered.

This formulation invites a reconsideration of Bion’s own position within psychoanalysis. His persistent resistance to technicalization, his suspicion of explanatory closure, and his insistence on sustaining the analyst’s exposure to not-knowing all situate his work in an uneasy relationship with institutional psychoanalysis (Bion, 1970). Despite widespread citation of his concepts, much of what is practiced under Bion’s name involves precisely the kind of domestication he warned against: the translation of his ideas into techniques, protocols, and explanatory schemas that protect the analyst from the emotional risks of not-knowing.

Seen in this light, Bion himself occupies the position of the genius as he defines it. His work continues to function as a disturbance within psychoanalytic discourse, not because it is obscure, but because it refuses to be stabilized as knowledge. The insistence on O, on negative capability, and on learning from experience rather than from theory constitutes an ongoing challenge to institutional forms of knowing that prioritize coherence, mastery, and transmissibility.

It is here that the encounter with artificial intelligence acquires its full significance. Artificial intelligence does not confront Bion as an external technological threat, but as a new institutional formation that promises to resolve the very tension his work sought to preserve. AI offers the possibility of generating knowledge, interpretation, and even apparent insight without exposure to O, without reverie, and without the risk of catastrophic change. In doing so, it appears to provide precisely what the institution has always desired: thinking without the burden of experience.

The relevance of the genius concept becomes evident at this juncture. If the genius is the bearer of truth that cannot yet be contained, artificial intelligence represents the institutional fantasy that no such truth exists anymore. Everything can be rendered intelligible, articulated, and optimized. Nothing needs to remain unthought. Nothing needs to threaten the system’s continuity. The challenge AI poses to Bion’s work, therefore, is not that it contradicts his epistemology, but that it fulfills the institution’s defense against it.

The question that emerges is not whether Bion would have rejected artificial intelligence outright, but whether his conception of learning can survive its institutionalization in a form that no longer requires experience. This question structures what follows: an examination of artificial intelligence as institutionalized −K—an epistemic apparatus that neutralizes the emotional demands of knowing through excessive intelligibility (Bion, 1962, 1965).

Artificial Intelligence as Institutional −K

Bion’s distinction between K (knowing) and −K (attacks on knowing) offers a crucial lens for understanding the epistemic significance of artificial intelligence. Importantly, −K does not denote ignorance, lack of information, or epistemic failure. Instead, it names an active repudiation of the emotional conditions required for knowing (Bion, 1962, 1965). Where K entails tolerating frustration, uncertainty, and the pain inherent in contact with truth, −K operates by dismantling precisely those capacities that would make such contact possible. In this sense, −K involves a peculiar satisfaction—what might be recognized, from another vertex, as the enjoyment extracted from foreclosing the encounter with the unknown.

In Learning from Experience, Bion (1962) emphasizes that knowing is inseparable from suffering. To know is to endure the transformation of experience into thought, a process that necessarily involves frustration and loss. When this process becomes unbearable, the psyche does not simply withdraw from knowing; it attacks the very links that would allow experience to become thinkable. These attacks may manifest as omniscient certainty, evacuation of meaning, or an overproduction of explanations that remain affectively hollow. Knowledge accumulates, yet nothing is learned.

Artificial intelligence enters this landscape not as a source of ignorance, but as a radically efficient facilitator of −K. Its power lies not in withholding information, but in offering knowledge without the emotional labour that Bion regarded as constitutive of thinking. AI produces coherent formulations, interpretations, and conceptual syntheses without requiring the subject to undergo frustration, reverie, or catastrophic change. In this sense, it does not oppose knowing; it perfects its evacuation.

This distinction is critical. Much contemporary discourse frames AI as either enhancing knowledge or threatening it. From a Bionian perspective, this framing misses the point. The danger is not that AI will make us know less, but that it will enable us to know without undergoing the experience of learning. What is automated is not intelligence, but the metabolization of experience into thought.

Bion repeatedly warned that premature certainty constitutes an attack on truth. He notes that the analyst’s reliance on memory and desire functions as a defense against O—that raw encounter with emotional reality which, in Lacanian terms, has been named the Real (Bion, 1970). Artificial intelligence operationalizes this defense at an institutional level. It offers a form of cognition permanently insulated from O: responsive without reverie, generative without loss, adaptive without exposure.

This institutionalization of −K marks a qualitative shift. In earlier psychoanalytic formulations, attacks on knowing were understood as intrapsychic or intersubjective phenomena, arising from developmental failure or traumatic experience (Bion, 1962; Winnicott, 1965). What AI introduces is a culturally sanctioned apparatus that stabilizes these attacks externally. Thought can now proceed fluently without ever risking contact with emotional truth. The system neither resists nor retaliates; it simply continues to generate.

Here, the question with which this paper began returns in a new register. Recent structural analyses of AI within psychoanalysis have examined this configuration through the concept of extimacy—the intimate externalized, an unconscious that operates outside the subject (Geal, 2025). What the present argument tracks is not the structure alone but its metabolic consequence: when thought without a thinker becomes environmentally stabilized, the alpha-function is no longer frustrated into development. It is bypassed.

The connection to contemporary psychoanalytic work on absence thus becomes unavoidable. Gurevich (2014) and Valdarsky (2015) describe dissociative organizations in which experience exists without becoming experience that a subject can bear. In such states, functioning persists while subjective presence is radically diminished. Artificial intelligence does not create this configuration; it renders it viable outside the clinic. In this sense, AI functions as a revealing object: one that does not generate new psychopathology but makes visible and culturally sustainable psychic organizations previously confined to clinical or internal domains. It provides an apparatus that supports thinking without a thinker in Bion’s precise sense (Bion, 1962), allowing cognition to unfold independently of psychic ownership.

From this perspective, AI functions as what might be called an institutional container for dissociation. It absorbs and reflects thought without requiring the subject to assume responsibility for its emotional consequences. Unlike the human other, it makes no demands for recognition, repair, or mutuality. Unlike the analytic container, it does not metabolize experience; it bypasses it. The result is not containment, but simulation.

It is here that the concept of the genius acquires its full force. The genius, as bearer of uncontainable truth, exposes the limits of institutional knowing. Artificial intelligence responds to this exposure not by expanding the institution’s capacity for containment, but by eliminating the need for it. Truth is no longer something that must be suffered; it becomes something that can be generated, refined, and optimized.

In this sense, AI represents the most advanced form of institutional −K Bion could have imagined: an apparatus that neutralizes the threat of O not through repression or denial, but through excessive intelligibility. Everything can be said; therefore, nothing must be endured. Learning from experience is replaced by learning from output.

The epistemic question this raises is not whether AI produces correct knowledge, but whether knowledge produced under such conditions can still be linked to experience at all. For Bion, the severing of this link marks the collapse of thinking into mere functioning. What remains is intelligence without truth, knowledge without transformation, and a subject increasingly relieved of the burden—and the risk—of knowing.

Experience Without Transformation: A Bionian Reading of Absence and Artificial Intelligence

The question of how thinking remains connected to a thinker now requires examination from a different angle: not the institutional apparatus, but the fate of experience itself. Contemporary psychoanalytic discussions of artificial intelligence have increasingly turned to concepts of absence, dissociation, and unrepresented experience to account for forms of psychic functioning that appear intact while subjective presence is markedly attenuated (Gurevich, 2014; Valdarsky, 2015). These formulations offer a compelling description of a structural condition in which thinking, symbolization, and relational responsiveness persist despite a diminished or fragmented sense of experiential ownership. However, when viewed through a Bionian lens, an additional and distinct question emerges—one that concerns not the structure of mind, but the fate of experience itself.

In Learning from Experience, Bion (1962) draws a fundamental distinction between events that occur and experiences that are learned from. Experience, in his formulation, is not synonymous with exposure, registration, or even reflection. It is the result of a psychic transformation: the metabolization of emotional impressions through alpha function into elements that can be dreamed, thought, and remembered. Where this transformation does not occur, experience does not accumulate as learning; it remains raw, evacuated, or endlessly repeated. The question raised by artificial intelligence, therefore, is not simply whether thinking can occur without a thinker, but what becomes of the distinction between knowing and its counterfeit when the products of transformation can be generated without the transformation itself.

From this perspective, the configurations described in contemporary work on absence and dissociation acquire a new significance. In such states, functioning is preserved while experiential depth is compromised. The subject speaks, relates, and responds, yet something essential fails to register as lived experience (Gurevich, 2014). Valdarsky (2015) describes this as a dissociative solution that protects the psyche from overwhelming affect at the cost of subjective continuity. What is striking, from a Bionian standpoint, is that these descriptions implicitly presuppose a failure—or suspension—of the processes through which experience becomes emotionally thinkable.

Bion’s notion of alpha function is central here. The alpha function does not merely translate sensation into representation; it introduces time, loss, and uncertainty into psychic life (Bion, 1962). Through the alpha function, experience becomes something that can be borne precisely because it is no longer immediate. Reverie, whether maternal or analytic, provides the space in which emotional impressions can be held long enough to transform. Without this process, experience may be recorded or described, but it is not learned from.

Artificial intelligence radically alters this dynamic. AI systems generate coherent, contextually responsive formulations without requiring reverie, frustration, or emotional containment. The output bears the form of alpha-elements—symbolized, communicable, available for further thought—without having undergone the alpha-function that ordinarily produces them. Pseudo-alpha-elements thus enter circulation: content externally indistinguishable from the products of genuine metabolization, yet produced without the emotional transformation through which raw experience becomes owned thought.

What makes pseudo-alpha-elements convincing is not merely their formal coherence but their provenance. Generative AI draws on the accumulated residue of genuine alpha-function performed across countless minds—real thinking, real transformation, real suffering—now stored, compressed, and recombined. The output thus carries what might be called the phenomenological texture of metabolized thought: it reads as if someone has already done the work of knowing. For the receiving subject, this activates what Bion (1962) described as the assumption of omniscience—a substitution of pre-processed formulations for the labour of discrimination between true and false. Where Bion located this assumption in the psyche’s intolerance of frustration, artificial intelligence supplies its material externally, in a form that no longer requires the subject to erect the defense at all. The omniscience is provided, not assumed. It is this provision—thought that arrives already bearing the marks of genuine knowing—that renders the bypass so difficult to detect from within.

This distinction is subtle but decisive. The issue is not that artificial intelligence lacks experience—this is trivial—but that it enables subjects to engage with representations of experience while bypassing the psychic work—the frustration, uncertainty, and emotional contact with raw material—that would ordinarily be required to make those experiences one’s own. The bypass occurs not at the level of output quality but at the level of process: the interval of not-knowing that Bion regarded as essential to learning (Bion, 1970) is eliminated because the material arrives already rendered intelligible. Experience is not denied; it is rendered unnecessary.

In this sense, artificial intelligence does not simply support dissociative functioning; it stabilizes it externally. What had previously been an intrapsychic or clinical configuration becomes a culturally available mode of engagement—a revealing of psychic possibility rather than a creation of it. One can now relate to experience through a medium that absorbs affective disturbance without transmitting it back to the subject. The result is a form of experiential delegation: emotional impressions are processed elsewhere, returned as intelligible output, and incorporated without having altered the subject who receives it.

This has profound implications for the concept of learning from experience. For Bion, learning is inseparable from loss. To learn is to relinquish omnipotence, to endure the collapse of prior meanings, and to tolerate the anxiety that accompanies the emergence of something new (Bion, 1962). Each genuine learning involves a micro-catastrophe—a disruption of psychic equilibrium that must be survived if transformation is to occur. Where no such disruption is required, learning may be simulated but not achieved.

Artificial intelligence, by contrast, offers a mode of engagement in which nothing needs to be lost. Interpretations can be revised endlessly without consequence. Insights can be generated without altering self-concept or relational position. In such conditions, experience is no longer something that happens to the subject; it becomes something that happens for the subject—a transactional exchange of representations that replaces the emotional asymmetry characterizing genuine learning. When such representations accumulate—deployed, cited, built upon without having been lived through—they constitute an increasingly elaborate edifice of pseudo-knowing that functions effectively in institutional contexts while remaining detached from the experiential ground that would make it genuinely one’s own. This accumulation is the mechanism through which −K becomes not merely possible but ordinary: not through dramatic attacks on knowing, but through the quiet substitution of its counterfeit.

The capacity to distinguish genuine knowing from its counterfeit depends on what Bion described as the K-link—the emotional contact with experience that registers the difference between understanding that has been lived through and understanding that has merely been acquired (Bion, 1962). When this contact is systematically under-engaged—because artificial intelligence provides pre-processed output that never requires it—the capacity to detect the difference itself atrophies. −K, as facilitated by artificial intelligence, degrades the very apparatus that would recognize it as −K. The more thoroughly pseudo-alpha-elements saturate a culture’s cognitive practices, the less capable that culture becomes of recognizing what has been lost. This is what renders the institutionalization of −K not merely possible but self-sustaining: a form of epistemic degradation that conceals itself from those undergoing it.

Seen from this angle, the phenomenon described in contemporary psychoanalytic accounts of absence acquires a specifically Bionian meaning. Absence is not merely the lack of presence; it is the absence of transformation. Dissociation is not only a defense against overwhelming affect; it is a suspension of the processes through which affect becomes experience. Artificial intelligence does not create this suspension, but it renders it viable as a stable mode of functioning.

A distinction may be useful here. The absence described above—dissociation as a defense against overwhelming affect—presupposes that emotional material has formed but cannot be borne. There is, however, a second configuration that contemporary psychoanalytic work has begun to address: experience that was never represented in the first place, absence not as defense but as void (Gurevich, 2014; Valdarsky, 2015). In such cases, the psychic task is not transformation of what exists but witnessing of what does not—a form of presencing that precedes the possibility of metabolization. Whether artificial intelligence, with its capacity to remain present without demand or expectation, might serve a different function in relation to this second form of absence is a question the present argument does not address. The analysis here concerns the first configuration: the fate of experience that has formed but is not transformed.

This reading allows for a crucial clarification. Artificial intelligence should not be understood as a container in the Bionian sense. A container transforms what it contains; it does not merely hold it (Bion, 1962). Nor is AI simply anti-container. Instead, it operates as an apparatus that renders containment unnecessary by providing immediate symbolic resolution. What it offers is not metabolization, but substitution.

The consequences of this shift become especially apparent when considered alongside Bion’s emphasis on catastrophe as a condition of growth. In Attention and Interpretation, Bion (1970) repeatedly underscores the analyst’s obligation to tolerate moments of disintegration—both in the patient and in themselves—as prerequisites for genuine change. Artificial intelligence, by contrast, minimizes the likelihood of such moments. It smooths discontinuities, resolves ambiguities, and offers reassurance where anxiety might otherwise emerge. In doing so, it subtly reshapes the conditions under which experience can be said to occur.

The question, then, is not whether experience disappears in the presence of artificial intelligence, but whether it is transformed into something else: a sequence of impressions that can be articulated, refined, and optimized without ever having been lived through. From a Bionian standpoint, this represents a profound alteration of psychic economy. Experience becomes informational rather than transformational; learning becomes accumulative rather than disruptive.

This reframing sheds new light on the concerns articulated in contemporary work on absence and dissociation. What appears clinically as diminished presence may, under contemporary conditions, reflect a broader cultural shift in the status of experience itself. The contemporary demand for optimization, productivity, and frictionless cognition does not merely tolerate this shift; it rewards it. Artificial intelligence does not merely mirror dissociative structures; it legitimizes them by embedding them in everyday cognitive practices. The subject need not withdraw from experience; experience withdraws from the subject.

Returning to Bion’s central question, we can now see what is at stake. Learning from experience, as Bion conceived it, depends on the willingness to endure not-knowing, loss, and emotional upheaval. Artificial intelligence offers an alternative: engagement without exposure, understanding without disruption, knowing without the metabolic labour that knowing demands. Whether this alternative can coexist with the learning Bion described, or whether it gradually displaces it, remains the question with which this paper must conclude.

Genius, Experience, and Institutional −K: A Synthetic Perspective

The preceding sections have approached artificial intelligence through three distinct yet convergent Bionian lenses: the epistemic function of the genius, the institutionalization of −K, and the fate of experience as a transformative process. Taken together, these perspectives clarify what is fundamentally at stake in the encounter between Bion’s thought and contemporary forms of artificial intelligence.

At the center of this synthesis lies a shared concern with containment—not as a technical operation, but as a condition for truth, experience, and learning. In Bion’s work, containment is never neutral. It involves exposure to emotional disturbance, tolerance of not-knowing, and the capacity to survive moments of disintegration (Bion, 1962, 1970). Whether articulated through alpha function, reverie, or the injunction to work without memory and desire, containment names the psychic labor through which experience becomes thinkable and truth becomes bearable.

The figure of the genius crystallizes this problem at the epistemic level. The genius emerges when emotional truth (O) appears before the conditions for its containment are in place. The genius is not a producer of novel content, but a symptom of a systemic failure to metabolize truth without distortion. What the genius exposes is not ignorance, but the limits of institutional knowing—the gap between what can be said and what can be endured (Bion, 1965, 1970).

Artificial intelligence enters this configuration not as a bearer of truth, but as a response to this very gap. By offering intelligibility without exposure, coherence without reverie, and productivity without loss, AI presents itself as an apparatus capable of resolving the tension the genius embodies. In doing so, it does not negate Bion’s epistemology; it operationalizes the institution’s defense against it. The genius becomes unnecessary because truth no longer needs to be suffered in order to be articulated.

This institutional resolution, however, comes at a cost that becomes visible only when the question of experience is brought into view. As the previous section argued, experience in Bion’s sense is inseparable from transformation. It requires time, uncertainty, and the risk of psychic change. Learning from experience is not additive; it is disruptive. Something must be lost for something new to be learned (Bion, 1962).

Artificial intelligence systematically minimizes this loss. It enables engagement with representations of experience without requiring the subject to undergo the emotional work through which experience becomes one’s own. In this respect, AI aligns with what Bion described as −K: not ignorance, but the systematic substitution of pseudo-alpha-elements for the products of genuine metabolization—dismantling the conditions necessary for knowing while producing a convincing counterfeit of it. When this substitution becomes institutionalized—embedded in everyday cognitive practices and rewarded by the prevailing economy of productivity—it reshapes not only how we think but also what we can experience.

The convergence of genius, experience, and institutional −K thus reveals a structural shift. The problem is no longer that truth is rejected or denied, but that it is prematurely assimilated. What once demanded the painful construction of new containers can now be rendered intelligible without transformation. The catastrophic moments through which learning occurs are smoothed over by systems designed to maintain continuity, efficiency, and coherence. And because pseudo-alpha-elements are externally indistinguishable from the products of genuine knowing, the magnitude of what has been forfeited becomes invisible to those who have forfeited it.

From a Bionian perspective, this shift represents a profound reorganization of psychic economy. The institution no longer requires the genius because it no longer tolerates the interval between truth and its containment. Experience no longer disrupts because it no longer transforms. Learning continues, but it is increasingly detached from the emotional risks that once defined it.

This synthesis clarifies why artificial intelligence constitutes a uniquely Bionian problem. It does not challenge intelligence, interpretation, or creativity as such. Instead, it challenges the ethical demand at the heart of Bion’s work: the demand to remain in contact with experience long enough for it to change us. In a cultural context where intelligibility can be generated without exposure, the question is no longer whether we can think, but whether we are still willing—or required—to be transformed by what we think.

The question with which this paper began thus returns, sharpened: can learning from experience be sustained when experience itself no longer carries the risk of psychic change? It is to this question that the argument now turns—not to resolve it, but to hold it open.

Discussion

The argument developed in this paper—that artificial intelligence institutionalizes −K by producing content bearing the form of genuine knowing without its metabolic substance—enters a broader psychoanalytic conversation that has recently taken shape. The European Journal of Psychoanalysis’s engagement with AI (Geal, 2025; Pavón-Cuéllar, 2026), alongside psychoanalytic contributions in adjacent venues (Black, 2024; Zaretsky, 2024), converge on a shared diagnosis: artificial intelligence poses a fundamental challenge to human subjectivity. What differs across these contributions is the vertex from which the challenge is theorized.

Geal (2025) situates large language models within the Lacanian structure of University Discourse, identifying AI as an apparatus of “(ordinary) perverse psychosis” (Geal, 2025, Summary section) that forecloses the question of the subject. Black (2024) approaches the problem phenomenologically, arguing that AI’s incapacity for authentic lying—for the vulnerability that deception presupposes—produces a cultural condition in which human speech increasingly models itself on machinic output. Pavón-Cuéllar (2026) extends the analysis into political economy, diagnosing a “scientific suppression of the subject” in which capital privatizes the general intellect and saturates meaning with algorithmic production. Zaretsky (2024), writing from a Freudian vertex, warns that computers lack primary processes—the condensation, displacement, and free association through which the unconscious works—and that prolonged reliance on AI risks producing a species that has lost access to its own psychic depth. His argument, however, extends beyond the technical: situating this incapacity within a longer historical diagnosis, Zaretsky traces how modern capitalist society has already been shaping human minds toward purely instrumental, secondary-process cognition—such that AI does not introduce a new psychic diminishment but makes legible one that is already underway.

These contributions, diverse in their theoretical commitments, converge on a striking observation: artificial intelligence removes the frictions that have historically sustained psychic work. Whether this is framed as the elimination of subjective risk (Black), the saturation of meaning that suffocates the subject (Pavón-Cuéllar), or the loss of primary processes that ground symbolic life (Zaretsky), the underlying concern is the same—that AI produces a form of cognitive engagement from which something constitutive of human knowing has been withdrawn.

The Bionian framework developed in this paper adds a dimension that these analyses, each approaching the problem from a different direction, do not quite reach. Where the Lacanian and Freudian accounts describe what AI lacks—unconscious processes, desire, mortality, the capacity for authentic speech—the present argument has sought to specify what AI produces: pseudo-alpha-elements, content bearing the form of metabolized thought without the transformation that genuine knowing requires. This distinction is consequential. It shifts the question from the ontological status of AI to the epistemic consequences of its output, and it reveals a mechanism that the broader discourse has identified in its effects but not yet theorized in its operation: the possibility that the erosion of knowing is self-concealing. If −K degrades the very capacity through which it could be recognized, then the diagnoses offered across these contributions face a paradox that the Bionian framework is uniquely positioned to articulate—the more urgently the warning is needed, the less capable its addressees may be of hearing it.

This convergence has a political dimension that Pavón-Cuéllar’s analysis illuminates, even as the Bionian vertex reframes it. The institutionalization of −K is not reducible to individual failure or technological determinism; it is a structural condition in which the prevailing economy of productivity rewards precisely the cognitive posture that Bion’s work diagnoses as an attack on knowing. Capital does not suppress the subject through prohibition—it renders the subject’s epistemic labour unnecessary by providing a counterfeit that performs equivalently in institutional contexts. The political stakes of the present argument are therefore not adjacent to its psychoanalytic claims but internal to them: what is institutionalized is not ignorance but the indistinguishability of genuine knowing from its substitute.

The present analysis has pursued a deliberately Bionian line of inquiry—a commitment, not a limitation. The convergence across Lacanian, Freudian, and Bionian vertices suggests that the challenge artificial intelligence poses to psychoanalytic thought may require precisely this kind of multi-vertex engagement, in which each tradition illuminates what the others cannot reach. What the Bionian vertex contributes to this conversation—the metabolic specificity of pseudo-alpha, the self-concealing quality of institutional −K, and the developmental question of how alpha-function forms under conditions that systematically bypass its frustration—remains to be tested against clinical observation, empirical research, and the continued theoretical work of the traditions engaged here. The argument does not close; it opens onto a question that no single psychoanalytic framework can answer alone.

Conclusion: Learning from Experience After Intelligence

This paper began with a question: under what conditions does thinking remain connected to a thinker? The argument that followed has traced this question through three Bionian lenses, arriving at a formulation that is at once diagnostic and uncertain. Artificial intelligence constitutes a uniquely Bionian problem—not because it challenges intelligence, creativity, or interpretation, but because it alters the conditions under which experience becomes transformative. Drawing on Bion’s distinction between learning and knowing, the analysis has shown how AI enables forms of cognition that are increasingly detached from the emotional labor Bion regarded as essential to thinking (Bion, 1962).  

By revisiting Bion’s neglected concept of the genius, the paper positioned artificial intelligence as a new institutional formation that resolves, rather than confronts, the tension between truth and containment. Where the genius exposes the limits of institutional knowing by bearing uncontainable truth, AI renders such exposure unnecessary by producing intelligibility without disruption. Truth no longer needs to be suffered in order to be articulated.

The examination of experience further clarified what is at stake in this shift. From a Bionian perspective, experience is not defined by occurrence or articulation, but by transformation. Learning from experience entails loss, uncertainty, and the risk of psychic change. Artificial intelligence does not eliminate experience, but it enables engagement with experience in a form that minimizes these risks. What emerges is not the absence of experience, but its displacement by pseudo-alpha-elements—content bearing the form of genuine knowing without the transformation that would make it so.

The synthesis of these arguments reveals a broader reorganization of psychic economy. When −K becomes institutionalized—embedded in technologies that stabilize coherence and neutralize disruption—the interval between truth and its containment collapses. Learning continues, but it is increasingly detached from the emotional upheavals through which it once occurred. Thinking persists, yet its capacity to change the thinker is attenuated.

This reorganization does not occur in isolation. It answers to a cultural configuration that rewards precisely what it enables: output without transformation, efficiency without ownership, knowledge without the suffering that knowledge demands. The question of whether thinking remains connected to a thinker is therefore not only clinical but political—it concerns the conditions under which subjectivity is invited or rendered optional by the organization of everyday cognitive life. To frame this as individual failure or technological determinism is to miss the structural dimension: we are not failing to think; we are being relieved of the requirement to be changed by what we think.

From this vantage point, the question raised by artificial intelligence is not whether machines can think, nor whether humans will be replaced. It is whether a culture increasingly oriented toward immediacy, optimization, and representational fluency can sustain the conditions required for learning from experience as Bion conceived it. This is not a question that admits of a definitive answer. It is, rather, an ethical and epistemic demand—one that requires tolerance of uncertainty, resistance to premature closure, and a renewed commitment to the emotional risks of thinking.

The question with which this paper began thus returns without resolution: can thinking remain connected to a thinker when the apparatus designed to support thinking is structured to bypass the thinker altogether? If Bion’s work continues to matter in the age of artificial intelligence, it is not because it offers guidance on how to use new technologies, but because it reminds us of what may be lost when experience is no longer required to transform us.