The AI pseudo-encounter is the interaction with a quasi-personal system in which the sensation of being heard is real on the human side, but the being-heard itself does not take place, because the system is no second personal pole. It is ontologically no encounter — but its psychologically effective surrogate.
Structural Features
A real encounter demands two persons who are each directed toward one another out of their own original intentionality, who open themselves to one another in self-transcendence, and who span a shared space of meaning. In the pseudo-encounter the second side is missing:
- No Thou-pole. The “answer” of the AI comes from a system that is no person (Spaemann: something vs. someone).
- No self-transcendence on the other side. The system does not transcend itself; it generates tokens.
- No shared form of life. What Wittgenstein named as the condition of understanding (PI §241) is missing — the AI has no form of life in which its language would be embedded.
What remains is the effectiveness of the human pole: he feels heard, he unburdens himself in speaking, he experiences something that very much resembles an encounter. That is phenomenologically real and psychotherapeutically usable — but ontologically remains a mirroring.
Shannon Vallor’s Image
Shannon Vallor formulates it pointedly in The AI Mirror (Oxford 2024): an LLM is a backward-looking mirror of collective language use. It contains no counterpart, but our own linguistic shape, statistically condensed. Whoever “speaks” with it speaks into a mirroring — and experiences himself returned in good formulation.
This diagnosis is not polemical: a mirror is a useful thing. But mirroring forms no virtues. Aristotle determines virtue as hexis, which arises through repeated encounter with the other (NE II). A mirror cannot train it.
Sherry Turkle’s Empirical Diagnosis
Sherry Turkle documents in Alone Together (2011), The Empathy Diaries (2021), and subsequent works a growing empirical pattern: permanent pseudo-encounter with commercial AI companion applications correlates with:
- decreasing tolerance for the friction of real encounter (the other person is not always available, not always benevolent, not always in the same mood)
- eroded conversational capacity in the sense of social improvisation
- substitution of the labor of understanding by the convenience of mirroring
Turkle calls this skill atrophy of social virtues: what is not practiced withers.
Why the Encounter Has No Face
For Emmanuel Levinas, the face of the other is ethically constitutive: it precedes every thematization, it obligates before I grasp it. The face is experience of the non-availability of the other. Exactly this non-availability is what the AI lacks: its output is arbitrarily resettable, regenerable, re-rollable. A simulated face is conceptually no face.
From this follows the Levinas-strict diagnosis: the pseudo-encounter is not only no face, it cannot become a face — not even through better language models, photorealistic avatars, or embodiment. The missing structural feature is not quantity of data, but personhood.
Differentiation
Pseudo-encounter is not bad through and through. Whoever, lonely, dictates a letter that never arrives; whoever talks alone with a dying person without expecting an answer; whoever presents to an AI complaints that are taboo in the family — all that has its meaning, as long as it remains clear what it is. Pseudo-encounter becomes problematic where it displaces real encounter and one’s own perceptual register for the Thou is unlearned.
Ontological Classification
- is disjoint from: encounter
- is a risk for: self-transcendence, communio personarum, the formation of personal virtue
- typically occurs in: AI conversation simulation with AI companions
- contains neither: face nor genuine address
Sources: Generated by querying the Personhood ontology.
Further sources:
- Turkle, Sherry (2011): Alone Together. Why We Expect More from Technology and Less from Each Other. New York: Basic Books.
- Turkle, Sherry (2021): The Empathy Diaries: A Memoir. New York: Penguin Press.
- Vallor, Shannon (2024): The AI Mirror. Oxford: Oxford University Press.
- Levinas, Emmanuel: Totality and Infinity, transl. Alphonso Lingis. Pittsburgh: Duquesne University Press, 1969 (French original 1961).
- Buber, Martin: I and Thou, transl. Ronald Gregor Smith. Edinburgh: T&T Clark, 1937 (German original 1923).
- Spaemann, Robert: Persons. The Difference between ‘Someone’ and ‘Something’, transl. Oliver O’Donovan. Oxford: Oxford University Press, 2006 (German original 1996).
- Aristotle: Nicomachean Ethics II (Bekker pagination). In: The Complete Works of Aristotle, ed. Jonathan Barnes, transl. W. D. Ross. Princeton: Princeton University Press, 1984.