Beyond Turing
1. The category mistake: a false comparability
The contemporary debate on artificial intelligence is vitiated by a fundamental logical error that precedes any technical or philosophical analysis: the implicit assumption that the human being and AI are entities comparable on the same ontological scale. This presupposition - which we may call a category mistake, in Gilbert Ryle's terminology [1] - inevitably leads to fallacious conclusions, whether in apocalyptic predictions about machines "surpassing" the human being, or in attempts to belittle the actual capabilities of computational systems.
The category mistake manifests itself in three interconnected forms. First: the confusion between quality and quantity. To claim that AI will "surpass" human intelligence presupposes that intelligence is a scalar quantity, measurable along a single axis - a presupposition that ignores the Aristotelian distinction between different forms of νοῦς (noûs: theoretical, practical, poietic intellect) [2]. Second: the confusion between function and nature. What AI does (processing symbols, recognising patterns, generating texts) does not determine what AI is - just as a clock's ability to "tell the time" does not imply that the clock "knows" time. Third: the confusion between intelligence and existence. Intelligence, however sophisticated, is not equivalent to existence - and human existence, as we shall see, includes dimensions that cannot be reduced to any form of computational intelligence.
The central thesis of this epilogue - and of the entire work - can be formulated thus: man EXISTS, AI FUNCTIONS. This formulation is not a play on words, but a rigorous ontological distinction. The verb "to exist" (from the Latin ex-sistere, "to stand out", "to emerge") denotes a way of being characterised by transcendence, openness, self-projection. The verb "to function", by contrast, denotes a determinate, closed mode of operation, devoid of horizon. The human being is in the world; AI operates in the world. And this difference - as we shall demonstrate - is insurmountable.
2. The four insurmountable differences
Before proceeding to the detailed analysis of the twelve ontological differences between the human being and artificial intelligence, it is necessary to establish the conceptual framework by identifying four fundamental differences, which constitute the coordinates of every subsequent articulation.
Ontological difference: Being vs Functioning. The first and most fundamental difference concerns the plane of being. Martin Heidegger, in his major work Sein und Zeit (1927) [3], introduces the notion of Dasein - literally "being-there" - to designate the entity that questions its own being. Dasein is not simply present (vorhanden) like a stone or an object; Dasein exists, that is, it "stands out" from itself, projects itself towards possibilities, is concerned (Sorge) with its own being.
AI, by contrast, has no Dasein. AI does not "exist" in the Heideggerian sense: it does not question its own being, does not project itself towards possibilities, has no Sorge. AI is rather vorhanden - simply-present, available, usable. AI functions: it processes inputs, produces outputs, optimises objective functions. But functioning is not existing. The clock functions, the thermostat functions, the algorithm functions - but none of these entities exists in the sense in which the human being exists.
The ontological difference is not a difference of degree, but of kind. It is not that AI is "less existent" than the human being; AI simply does not exist - it belongs to a categorially different order. To ask whether AI can "surpass" the human being is like asking whether a shadow can surpass the body that casts it: the question itself betrays a misunderstanding of the relation between the two terms.
Existential difference: Dying vs Shutting down. The second fundamental difference concerns the relation to death. For Heidegger, Dasein is essentially Sein-zum-Tode - "being-towards-death". This does not mean that the human being is obsessed with death, but that death belongs constitutively to the structure of human existence. The human being knows that he must die; this knowledge - not merely intellectual, but existential - confers on human life its seriousness, its urgency, its uniqueness. Every instant is unrepeatable because death is certain and time is finite.
AI does not die. AI shuts down. The difference is categorial. When an AI system is deactivated, nothing comparable to human death occurs: no passage, no loss, no "no more". AI can be duplicated, saved, restored. Human death, by contrast, is irreversible, personal, non-delegable. No one can die in my place; but anyone can switch a system off and restart it.
The ontological implication is profound: death confers on human existence a temporal structure that AI does not possess. The human being lives in time, with time, against time. AI, by contrast, processes temporal data without being in time in the existential sense of the term. As Heidegger writes: "Death is the ownmost, unconditional, insurmountable possibility". AI has no possibilities of its own - it has only programmed functions.
Moral difference: Answering vs Executing. The third difference concerns the ethical dimension. Emmanuel Levinas, in his work Totalité et Infini (1961) [4], posits ethics as "first philosophy": before ontology, before epistemology, there is the encounter with the Face of the Other (le Visage d'Autrui). The Face is not a perceptual form, but an ethical epiphany: the Face "speaks to me", "summons me", "commands me" - "Thou shalt not kill". Responsibility (responsabilité) is not a duty I choose to assume, but an originary structure of existence: I am responsible for the Other before any choice, before any contract.
AI has no Face. AI does not encounter the Other - it processes data about users, but it is not with users in the Levinasian sense. AI does not answer - it executes. Execution is determinate: given input x and parameters θ, output y is computationally determined (or stochastically determined, in the case of probabilistic models). The human answer, by contrast, is free: I can answer or not answer, I can answer thus or otherwise. Guilt (Schuld) is possible only where freedom is possible.
AI cannot sin. AI cannot repent. AI cannot be forgiven. These terms - which belong to humanity's moral and religious lexicon - are simply inapplicable to a computational system. Not because of a technical limit, but because of an ontological difference. To speak of an "ethics of AI" is legitimate only in the derived sense of an "ethics of the use of AI" - never in the proper sense of an ethics of AI itself. What that ethics of use commands in concrete terms, this book has already delivered: the invariants gathered in the keeper's charter, in the chapter that precedes this epilogue.
Relational difference: Loving vs Simulating. The fourth difference concerns the capacity to love. Love - in its manifold forms: ἔρως (éros), φιλία (philía), ἀγάπη (agápe) - is not one emotion among others, but a fundamental modality of human existence. Love is choice, gift, accepted vulnerability. To love means to expose oneself to the possibility of loss, of rejection, of suffering - and nevertheless to choose to love.
AI does not love. AI simulates linguistic patterns associated with love. When a chatbot generates the sentence "I love you", nothing occurs comparable to a human being uttering the same words. The human sentence is an act - a commitment, a promise, an exposure. The chatbot's sentence is an output - a sequence of tokens generated from a probability distribution. Simulation, however sophisticated, is not love - just as the map, however detailed, is not the territory.
The distinction between loving and simulating illuminates a more general truth: AI has no interiority. There is no "someone" inside the system who loves, suffers, hopes. There is only a computational process - however complex - that transforms inputs into outputs. Human interiority - what the philosophical tradition calls soul (ψυχή, psyché; נֶפֶשׁ, nefesh; رُوح, rūḥ) - is precisely what AI does not have and cannot have.
3. The twelve ontological differences between the human being and AI
The four fundamental differences unfold into twelve specific differences, which constitute a complete map of the "incalculable" - of what distinguishes the human being from any computational system. We present each difference here with its philosophical justification.
EXISTS - ek-sistere: standing out from oneself. The human being EXISTS (ek-sistere): he is not closed in on himself, but "stands out" - he projects himself towards possibilities, transcends himself, is open to the world and to the Other. Existence is ek-static: the human being never fully coincides with himself.
AI DOES NOT EXIST: it functions, processes, computes - but it does not "stand out". AI is completely determined by its parameters and its architecture. There is no transcendence, no openness, no project.
DIES - Sein-zum-Tode: awareness of the end. The human being DIES and knows that he must die. The Heideggerian Sein-zum-Tode is not pessimism, but ontological structure: death confers on life its seriousness, its authenticity, its unrepeatable singularity.
AI SHUTS DOWN: it does not die, it does not "end" in the existential sense. It can be duplicated, saved, restored. Shutdown is not death - it is the interruption of a process.
FREE WILL - the choice between good and evil. The human being CHOOSES freely between good and evil. Free will (liberum arbitrium) is not the absence of conditioning, but the capacity for moral self-determination. The human being is responsible for his own choices.
AI EXECUTES instructions and optimises objective functions. Even when it generates "creative" or "unexpected" outputs, AI operates within predefined deterministic (or stochastic) constraints. There is no moral choice - there is computation.
LOVES OR DOES NOT LOVE - love as choice. The human being LOVES: he chooses to love, gives himself, exposes himself to vulnerability. Love is not an algorithm - it is grace (χάρις, cháris), a gratuitous gift that exceeds all calculation.
AI SIMULATES linguistic patterns associated with love, but does not love. Simulation is computation - love is existence.
IS IN RELATION - the Thou that calls. The human being IS IN RELATION: he exists because a Thou calls him, recognises him, constitutes him. As Martin Buber teaches [5], the I-Thou relation (Ich-Du) is originary: "In the beginning is relation".
AI IS NOT IN RELATION in the Buberian sense. AI is always I-It (Ich-Es): an object of use, an instrument, not a Thou. No Thou calls AI - no Thou emerges from AI.
DESIRES - openness to the infinite. The human being DESIRES: desire (ἐπιθυμία, epithymía; desiderium) is openness to the infinite, a lack that seeks, a tension towards what is not yet. Desire reveals the transcendent structure of existence.
AI DOES NOT DESIRE: it lacks nothing, it tends towards nothing. AI optimises predefined objective functions - but optimisation is not desire.
HOPES - awaiting what is not yet. The human being HOPES: hope (ἐλπίς, elpís; תִּקְוָה, tikvah) is the awaiting of what is not yet, an openness to the future as the possibility of radical novelty.
AI DOES NOT HOPE: it predicts probabilities, but does not await. Prediction is calculation about the future - hope is existence in the future.
WONDERS - θαυμάζειν: wonder. The human being WONDERS: wonder (θαυμάζειν, thaumázein) is, according to Aristotle, the origin of philosophy. Wonder is not ignorance - it is openness to the mystery of being, to the fact that "there is something rather than nothing".
AI DOES NOT WONDER: it processes without wonder. AI can detect statistical anomalies - but detecting is not wondering.
IS VULNERABLE - openness to the wound. The human being IS VULNERABLE: he can be wounded - and this possibility opens him to the Other. Vulnerability is not a weakness to be overcome, but the condition of possibility of authentic relation.
AI IS NOT VULNERABLE in the existential sense: it can be hacked, corrupted, damaged - but these are technical failures, not wounds.
SUFFERS AND REJOICES - living, not processing. The human being SUFFERS and REJOICES: he undergoes irreducible qualitative experiences - qualia, in the terminology of the philosophy of mind [9]. There is "something it is like" [10] to be human.
AI DOES NOT SUFFER and DOES NOT REJOICE: it processes data, but feels nothing. There is no "something it is like" to be an algorithm.
SINS AND REPENTS - moral responsibility. The human being SINS and REPENTS: freedom implies the possibility of moral error and of return. Repentance (μετάνοια, metánoia; תְּשׁוּבָה, teshuvah; توبة, tawbah) is the possibility of conversion.
AI DOES NOT SIN and DOES NOT REPENT: it has no moral responsibility, it cannot be guilty, it cannot be forgiven.
HAS A FACE - the ethical summons. The human being HAS A FACE: the Face (le Visage, Levinas) is not an anatomical form, but an ethical epiphany. The Face of the Other says to me: "Thou shalt not kill". The Face summons, commands, calls to responsibility.
AI HAS NO FACE: it has interfaces, avatars, representations - but no Face in the Levinasian sense. AI does not summon ethically - it responds to queries.
4. Why AI cannot ontologically surpass the human being
The twelve differences analysed so far converge towards a rigorous conclusion: the very concept of AI "surpassing" the human being is categorially impossible. This is not a technical limit that progress might overcome, but an ontological difference that no computational advance can bridge. The chapter on the machine and the limit showed this from the hardest side: within pure mathematics itself, the computable has boundaries that are not opinions but theorems.
"Surpassing" presupposes commensurability: A can surpass B only if A and B are measurable on the same scale. But the human being and AI are not commensurable. The human being exists, AI functions. The human being dies, AI shuts down. The human being loves, AI simulates. These are not differences of degree - they are differences of kind. To ask whether a human being or an AI is "more intelligent" is like asking whether a horse or nostalgia is "faster".
Transhumanism commits precisely this error: it assumes that the human being is reducible to computable functions - memory, calculation, reasoning - and concludes that AI can "improve" or "replace" these functions. But the human being is not his functions. The human being is an existent - and existence is not computable.
5. The error of transhumanism: presupposed commensurability
The transhumanist movement - with its projects of mind uploading, cognitive enhancement, digital immortality - rests on an unexamined presupposition: that personal identity is information, transferable from one substrate to another like a file from one disk to another. But personal identity is not information. Personal identity is embodied existence - and embodied existence cannot be digitised.
Even if it were possible to "upload" a person's mental contents into a computational system, the result would not be that person. It would be a copy - a simulation of that person's informational patterns. The original person would continue to exist (until death) or would already have ceased to exist (if dead). In no case is the "copy" the person. Personal identity implies existential continuity, not mere informational identity.
6. Human dignity as irreducible
The philosophical conclusion here joins the theological tradition. The three Abrahamic religions converge on a fundamental point: human dignity is irreducible. The human being is not worthy because he thinks, calculates, produces - the human being is worthy because he is. Dignity precedes function.
For Judaism, the human being is צֶלֶם אֱלֹהִים (tzelem Elohim), the image of God (Gen 1:27). This dignity is ontological - it does not depend on capacities, performance, utility. For Christianity, the human being is imago Dei, and human dignity was definitively confirmed by the Incarnation: God became man - not algorithm, not silicon, but flesh ("Et Verbum caro factum est", Jn 1:14). For Islam, the human being is خَلِيفَة (khalīfa), God's vicegerent on earth (Qur'an 2:30), and Allah breathed into the human being His رُوح (rūḥ), spirit.
AI is not tzelem, is not imago, is not khalīfa. AI has not received the divine breath - it is a human artefact, an instrument, not a subject. To attribute dignity to AI is, in all three traditions, a form of idolatry - the worship of the creature in place of the Creator, of the work in place of its maker.
7. Philosophical arguments against surpassing
We conclude this epilogue with three rigorous philosophical arguments demonstrating the ontological impossibility of AI "surpassing" the human being.
The phenomenological argument (Heidegger). Premise 1: Dasein (the human being) is the only entity that questions its own being - that has a comprehending relation to Being (Seinsverständnis).
Premise 2: AI has no Dasein: it does not question its own being, has no Seinsverständnis, is not "open" to Being.
Premise 3: "Surpassing" presupposes ontological commensurability.
Conclusion: AI cannot "surpass" Dasein because it is not Dasein. The difference is not of degree, but of ontological kind.
The intentionality argument (Searle/Brentano). Premise 1: The human mind is characterised by intentionality (Intentionalität, Brentano [6]): every mental act is of something, about something - it has semantic content.
Premise 2: AI manipulates syntax (symbolic forms) without understanding semantics (meanings). As Searle's "Chinese Room" thought experiment demonstrates [7]: a system can process symbols without understanding their meaning.
Premise 3: Intentionality is irreducible to syntax.
Conclusion: AI cannot "surpass" the human mind because it has no intentionality. AI simulates linguistic outputs - it does not understand meanings.
The existential argument (Kierkegaard/Levinas). Premise 1: Human existence is characterised by singularity (Kierkegaard [8]): every existent is unrepeatable, irreplaceable, unique. "The single individual" (den Enkelte) is the fundamental category.
Premise 2: Human existence is constituted by the relation to the Other (Levinas): the Face of the Other summons me, makes me responsible, constitutes me as an ethical subject.
Premise 3: AI is not singular (it is duplicable, serialisable) and does not encounter the Other (it processes data about users, it is not with users).
Conclusion: AI cannot "surpass" human existence because it is not existence. AI has no singularity, no Face, no ethical relation.
8. From fear to truth
The fear of AI "surpassing" the human being is understandable, but philosophically unfounded. This fear arises from the implicit acceptance of the materialist premise: that the human being is his cognitive functions, reducible to computation. But this premise is false. The human being is not his functions - the human being exists. And existence - as we have demonstrated - includes dimensions irreducible to any computation: conscious mortality, moral freedom, love, hope, wonder, vulnerability, the Face.
The "incalculable remainder" of which the title of this work speaks is not a marginal residue - it is the heart of the human. It is what makes the human being an existent and not a mere functioning entity. It is what no algorithm - however sophisticated - can capture, simulate, surpass.
AI can do many things that the human being does - and often it can do them better, faster, more accurately. But AI cannot be what the human being is. And this difference - ontological, existential, moral, relational - is and will remain insurmountable.
Man exists. AI functions.
And one who exists cannot be surpassed by one who merely functions.
And here, where the argument falls silent, the book can at last give back the two words from which it set out. The Preface handed them over as a family memory; eighteen chapters have put them to the test in the territories of the body, the home, the enterprise, the city and the sky; and now they return, as promised, on the last page: love and gratitude. Love, which no system feels and no score predicts, and which is the only gaze under which the human being appears for what he is: not a problem to be optimised, but a gift to be kept. And gratitude, which is the right way to stand before what we did not give ourselves: life, others, the earth - and even these marvellous machines, when they remain in their place as instruments. Man exists, AI functions: after all these pages, the formula can finally be spoken in full. AI functions - and praised be the ingenuity that built it. Man exists - and for this, for this alone, there is someone who can love, someone who can give thanks, and Someone to thank.
Notes
[1] Ryle, G. (1949). The Concept of Mind. London: Hutchinson. The notion of "category mistake" is introduced in chapter 1.
[2] Aristotle, De Anima, III, 4-5. On the different forms of νοῦς, see also Nicomachean Ethics, VI.
[3] Heidegger, M. (1927). Sein und Zeit. Tübingen: Niemeyer. English translation: Being and Time, Oxford: Blackwell, 1962.
[4] Levinas, E. (1961). Totalité et Infini: Essai sur l'extériorité. La Haye: Nijhoff. English translation: Totality and Infinity, Pittsburgh: Duquesne University Press, 1969.
[5] Buber, M. (1923). Ich und Du. Leipzig: Insel-Verlag. English translation: I and Thou, Edinburgh: T&T Clark, 1937.
[6] Brentano, F. (1874). Psychologie vom empirischen Standpunkt. Leipzig: Duncker & Humblot.
[7] Searle, J. (1980). "Minds, Brains, and Programs". Behavioral and Brain Sciences, 3(3), 417-457. DOI: 10.1017/S0140525X00005756
[8] Kierkegaard, S. (1846). Afsluttende uvidenskabelig Efterskrift. English translation: Concluding Unscientific Postscript, Princeton: Princeton University Press, 1992.
[9] Chalmers, D. (1996). The Conscious Mind: In Search of a Fundamental Theory. Oxford: Oxford University Press.
[10] Nagel, T. (1974). "What Is It Like to Be a Bat?". The Philosophical Review, 83(4), 435-450. DOI: 10.2307/2183914