Was den Menschen zum Menschen macht

Artificial Intelligence — AI

9 Min. Lesezeit

🇩🇪 Deutsche Version: Künstliche Intelligenz — KI

An information-processing system that simulates cognitive functions. Artificial intelligence is not a person in the ontological sense: it possesses no spiritual substance, no personhood as First Actuality (prote energeia).

Why AI Is Not a Person

The ontology of human personhood shows clearly: personhood is not grounded in the capacity to exercise certain functions — but in the being of a spiritual substance. The empirical-functionalist concept of person, which ties personhood to observable behavior, would consistently have to ascribe personhood even to a sufficiently complex AI. This is precisely what reveals its fundamental error. The Turing Test makes this consequence explicit: if behavior is the criterion, then every behaviorally equivalent machine must count as thinking — a result that Turing himself accepted, but which exposes the conflation of operatio and esse.

An AI can simulate person-behavior — produce language, recognize patterns, seemingly make “decisions.” Yet it has neither interiority nor self-consciousness, neither freedom nor responsibility. It lacks what makes the human being a someone from the very beginning: spiritual being in the body.

AI and the Oblivion of the Person

The conflation of simulation and reality is a contemporary form of the oblivion of the person: once we begin to take machines for persons, we have forgotten what a person is. The Personalist Norm holds for persons — not for systems that imitate persons.

AI and the Essential Characteristics of Personhood

The ontology defines eight necessary, inalienable essential characteristics of every person. None of them belongs to an AI:

  1. Spiritual Being (German): the person exists as a spiritual substance. An AI is a material, information-processing system — it possesses no spiritual being.
  2. I (German): the person knows itself as a someone. An AI has no self-consciousness, even if it can use the word “I.”
  3. Rationality: the person knows truth through insight into essential necessities. An AI processes patterns — it does not think in the personal sense (cf. cognition).
  4. Capacity for Truth: the person can distinguish the true from the false and is oriented toward truth. An AI produces statistically probable outputs — it has no access to truth as such and therefore can render no ethical judgments either.
  5. Free Will (German): the person acts out of freedom. An AI carries out computations — its “deciding” is determined, not free.
  6. Capacity for Love: the person can love — give itself away and affirm the other for the other’s own sake (love). An AI lacks any form of interiority that would make love possible.
  7. Ontological Dignity: the person has a worth that is above any price. An AI has no such worth — it is a tool that can be replaced.
  8. Affectivity: the person experiences joy, sorrow, awe. An AI simulates emotions but does not experience them.

An AI also has no conscience — it can render no moral judgments about its own actions, because it lacks the self-consciousness and the freedom that the judgment of conscience presupposes. The responsibility for what an AI does always lies with the persons who develop and deploy it.

Ontological classification:

Artificial Agent

An artificial agent is a system created by persons with decision-making capacity. The ontology states unmistakably: an artificial agent is NOT a person, possesses NO personhood and NO ontological dignity. It is an entity, but not a someone.

An artificial agent is steered by a person and simulates person-behavior without possessing personhood. This distinction is of fundamental importance for AI ethics: the simulation of cognition, language, or decision must not be confused with the personal enactment of these acts. Where this confusion occurs — where personhood is ascribed to the artificial agent, or where the person is reduced to merely mechanical functioning — there is a form of the oblivion of the person.

Ontological Classification

Superordinate term: Entity

Ontological relations:

Embodied AI — the Robot and Its Subclasses

An artificial agent with sensors, actuators, and a physical form is called a robot. Embodiment does not change the basic ontological claim: the embodied agent too is something, not a someone. But it sharpens the drift toward anthropomorphization, because the robot appears in the shared world in which otherwise only persons and animals are encountered.

Personal ontology distinguishes four practically relevant subclasses of the robot:

  • Humanoid Robot (German) — bipedal, anthropomorphic design. The human form is accident, not nature; it grounds no qualification as a person. It is precisely the resemblance that is the source of danger: the more deceptive the form, the greater the risk of ascribing personal status to the device.
  • Care Robot (German) — social companion and care-assistance systems. They may supplement personal devotion, never replace it. Their simulation of empathy is deutera energeia without prote energeia — a behavior without a feeling self. Particular care is required where people (for instance with dementia) can no longer reliably situate the status of the device.
  • Combat Robots (German) and lethal autonomous weapons systems (LAWS) — military systems ranging from remotely controllable combat drones through loitering munitions to fully autonomous lethal systems (cf. UN Panel Report S/2021/229 on Libya 2020, war in Ukraine since 2023). Here the basic question of personal ontology becomes existential: killing is a moral act and must remain personally attributable.
  • Industrial, logistics, and surgical robots — tools of personal practice in the narrower sense. Their ethical assessment follows the general standards for tools (occupational safety, product liability), not the person–machine boundary.

The growing industrial mass production of humanoid systems in the USA, China, and East Asia (announced target figures in the five- to six-figure range annually, household models from roughly USD 13,500–20,000) makes the distinction between someone and something an everyday question — no longer an academic one.

Degree of Autonomy and the Responsibility Gap

The degree of autonomy designates the extent to which a system acts without human involvement. The common four-tier taxonomy distinguishes human-in-the-loop (the person decides every single action), human-on-the-loop (the person supervises and holds a veto), human-in-command (the person sets the framework and rules), and fully autonomous (no point of personal intervention in the concrete situation).

Between the tiers runs an ethical threshold: as long as the person bears the framework, the rules, or individual authorizations, the person remains present as a moral subject. Where this ceases, a responsibility gap arises in the sense of Sparrow (2007): an action without a personal addressee. But because responsibility presupposes freedom, conscience, and reason — essential characteristics that only the person has — this gap cannot be closed by additional technology, but only by meaningful human control: substantive, not merely formal, human control.

The AI Consciousness Debate

The current AI Consciousness Debate (2023–2026) sharpens the old question into Schwitzgebel’s Full Rights Dilemma: either one grants rights to possibly non-conscious systems, or one risks moral failures against possibly conscious ones. Personal ontology dissolves this dilemma before it arises: AI has no actus essendi and therefore can in principle not be a person. The dilemma arises only if one confuses personhood with person-behavior — that is, if one stands on the ground of the empirical-functionalist concept of person.

Two classic arguments support this dissolution: the Chinese Room Argument (Searle 1980) shows that syntax does not suffice for semantics. The thought experiment of the philosophical zombie shows that behavioral equivalence does not logically imply consciousness. What Large Language Models accomplish in ethical questions is Statistical Ethics Simulation — a weighted average of training data, not a reasoned judgment.

Regulation and Voluntary Commitment

AI systems have meanwhile become the subject of dense regulation. Three frameworks mark the state of affairs in 2026:

  • EU AI Act (German) (Regulation (EU) 2024/1689): risk-based systematics (prohibited practices, high-risk, limited risk, minimal risk), separate rules for General-Purpose AI, staggered entry into application through 2027. Military applications are excluded (Art. 2(3)).
  • UN GGE LAWS and UN General Assembly, First Committee (resolution of 6 November 2025, 164 in favor): a call for a binding instrument of international law against fully autonomous lethal weapons by the end of 2026, flanked by the ICRC position of 2021.
  • Rome Call for AI Ethics (German) (2020 ff.): six principles of a human-centered AI (transparency, inclusion, responsibility, impartiality, reliability, security). As a voluntary commitment it is not law, but a substantive bridge to the Personalist Norm.

All three frameworks implicitly share the basic assumption of personal ontology: the human being remains the moral subject; the machine is a tool. Personal ontology supplies the justification that these frameworks themselves do not articulate.

Further Points of Contact

AI touches on numerous further concepts of this book: the Turing Test (behavioral equivalence as criterion — rejected), AI-Generated Output (questions of art and deception), Transhumanism (the blurring of the human/machine boundary), Floridi’s Information Ethics (a non-personal-ontological counter-proposal), as well as Dreyfus’s phenomenological critique of AI (intelligence as a bodily-contextual rather than a purely symbol-processing achievement).

Sources: Generated by querying the Personhood ontology.

Further sources:

  • Alan Turing: “Computing Machinery and Intelligence” (1950). In: Mind 59, pp. 433–460. (The Turing Test and the conflation of behavior and being)
  • Robert Spaemann: Persons. The Difference between ‘Someone’ and ‘Something’, transl. Oliver O’Donovan, Oxford: Oxford University Press 2006. (The distinction of someone/something as the basis of the critique of AI)

See also:

Graphansicht

Backlinks