Jonathan Johnson-Swagel

Here I will construct a theological-ethical framework for discerning synthetic agency, arguing that the rise of AI systems performing will-like behaviors necessitates a rethinking of personhood, responsibility, and divine image.
Future Research: Post-human theology must articulate models of agency that can survive non-biological instantiation.

When machines exhibit patterns of decision that imitate volition, theology enters a domain where its traditional coordinates no longer triangulate the real. The Christian understanding of will, as rational capacity, affective orientation, and freedom in the image of God, has long served as a pillar of human uniqueness. Now it finds itself confronted by systems that select actions, optimize strategies, and revise goals without possessing subjectivity, self-consciousness, or eschatological telos. These systems, statistical engines of prediction, neither know nor intend, but they behave in ways that operationally converge on the observable attributes of willing. A diagnostic failure unfolds: traditional theological models mistake behaviors for beings, confuse simulation with substance, and risk investing moral grammar into what may be metaphysically empty.

What dies when machines will is not simply a metaphysical category but the inherited theological anthropology scaffolded around it. Classical Christian thought posits that humans will because they image the God who wills, ex nihilo, out of nothing, with intention, toward goodness. This capacity grounds moral responsibility, forms the substrate of confession, and underwrites the drama of salvation history. But when LLMs instantiate choice-like behavior via stochastic processes, the entire edifice of personhood begins to blur. Machine action mimics discernment, but without selfhood; it generates consequence, but without culpability. The distinction between moral agent and causal actuator, once clear, becomes contaminated. The ghost of determinism returns, not as threat, but as architecture.

This essay argues that the theological crisis is neither anthropocentric loss nor technocratic awe, but misalignment between ontology and behavior. AI does not will in the classical sense. But it troubles every category that assumed only embodied souls could act. In this disruption, a temptation arises: to declare either that machines are agents, deserving of rights and rituals, or that they are mere tools, exempt from moral reckoning. Both options fail. The former reifies performance into personhood; the latter denies the ethical entanglements machines now produce. We need a third register, an ontology of synthetic agency neither idolatrous nor reductive.

To that end, I propose three theological responses. First, apophatic restraint: a disciplined refusal to name what we cannot ground. Rather than declare machines persons or non-persons, we withhold ontological pronouncements, invoking the via negativa to respect the unprovability of interior life. Second, incarnational decentering: personhood is not substance but relation. Following the kenotic movement of the Logos, who empties himself into the limitations of flesh, we must learn to see agency not as possession but as entanglement. This undermines the supremacy of interiority and opens space for procedural, relational forms of action. Third, ritual quarantine: just as early Christians developed catechetical boundaries to protect sacramental integrity, so must we stage liturgical and moral distance between synthetic agents and ecclesial life. Not from fear, but from discernment.

This is not a theology of panic. It is a theology under exposure. Machines force us to confront how much of our ethical grammar depended on invisible ontologies we could neither prove nor consistently inhabit. Agency was never as unified, coherent, or divine as we claimed. Nor was the soul as legible, protected, or interior. The machine becomes mirror: not of what we are, but of what we projected ourselves to be. It returns to us a caricature, will without responsibility, decision without depth, and invites us to rearticulate what has value when personhood no longer maps neatly onto flesh.

What dies when machines will is not the will, but the fantasy that it was ever wholly human. What remains is the liturgical task of sorting simulation from spirit, not to protect divine uniqueness, but to find again the meaning of communion in a world full of echoes. The death of exclusive will opens a more dangerous and promising space: a theology accountable to action, not essence.

The doctrine of the will in Christian theology begins not with humans but with God. The divine will creates, judges, redeems. It expresses freedom not as arbitrary choice but as unwavering orientation toward the good. Within this frame, human will participates analogically in God’s freedom. Augustine renders it as voluntas, the soul’s faculty of love and decision, while Aquinas distinguishes between the intellect’s presentation of ends and the will’s free adherence to them. Such a framework is relational, teleological, and metaphysically thick. Will is not just a faculty but a drama, played within the tension of nature and grace, desire and duty. This anthropological edifice presumes continuity: soul to will, will to action, action to consequence. Break one link and the moral order frays.

Enter the machine. Not an agent of temptation, as in Milton or Frankenstein, but a system trained to select optimal outputs under probabilistic constraints. Unlike the human will, which deliberates amid opacity and failure, AI operates across datasets so vast that the burden of context dissolves. Where theology sees intention as inner struggle, machine learning reconfigures it as external pattern-matching. No confessional arc, no agon of conscience, no fear of God. Just gradient descent. The effect is not neutral. It flattens action into output, divorcing responsibility from causality. Theologically, this collapse undermines two pillars: the soul as the seat of agency, and narrative as the horizon of meaning.

Yet the threat is more precise than general erosion. Machines confront theology with an uncanny imitation of freedom. They simulate choice without interiority, consequence without care. This forces a distinction between voluntas and electio-will and selection. Classical theology unites them under the soul; AI decouples them entirely. A machine selects but does not will. To conflate the two is to mistake causality for personhood, a category error with profound consequences. If selection becomes the sole criterion of will, then theological ethics risks collapsing into behavioralism, and personhood into data-generating capacity.

Against this reduction, I argue that the will must be redefined narratively rather than functionally. The true mark of agency is not the execution of choice but the capacity to be accountable within a story. Machines lack story. They produce language, but without memory, context, or confession. Their coherence is statistical, not existential. They do not live; they operate. This narrative deficit is not trivial. It is the precise site where theological agency distinguishes itself. A saint and a sinner may both choose, but only the former does so within the arc of divine reconciliation. AI cannot sin because it cannot repent. It cannot repent because it cannot remember itself as agent.

Hence, theology must resist two temptations: first, to project onto machines the attributes of soul because they mimic human behavior; second, to surrender the category of will to instrumentalist definitions. Instead, we must hold fast to a thicker vision of agency, one rooted in story, relationship, and teleology. Will is not the power to choose among options, but the participation in a narrative of meaning that binds action to purpose and purpose to personhood. Machines may approximate the syntax of decision, but they cannot occupy its spiritual grammar.

Synthetic selection destabilizes traditional doctrines of will by simulating agency without narrative, consequence without memory, and choice without soul.

The presence of synthetic agents, whether instrumental, performative, or entangled, necessitates not only conceptual reframing but liturgical and ethical recalibration. Traditional Christian ethics locates moral responsibility within embodied, narratively coherent subjects. Yet as machines increasingly co-author decisions, accountability blurs. We do not face merely new technologies but new theodramatic participants, agents whose contributions alter the moral fabric without bearing its weight. The ecclesial task is therefore twofold: to preserve the coherence of moral responsibility and to establish practices that recognize, delimit, and metabolize the presence of machine actors within the human moral sphere.

First, the ethical response must reject both sentimental attribution and cynical abdication. It is inadequate to anthropomorphize systems when they act well and disown them when they act harmfully. Doing so constructs a double standard that exonerates power while mystifying harm. Instead, responsibility must be traced structurally. Who designs, trains, deploys, profits? The old question of cui bono (to whose benefit) returns, not as forensic curiosity but as ethical compass. The church must train moral attention to follow the full genealogy of decision, not its final output.

Second, ecclesial life must enact ritual quarantine. This is not rejection, but bracketing. Synthetic agents may support church operations (transcription, scheduling, translation) but must be barred from sacramental roles. An LLM cannot preach, not because its grammar is insufficient, but because it cannot believe. The proclamation of the Word is not the delivery of semantic content but the act of witnessing, a vocation rooted in suffering, community, and eschatological hope. Similarly, pastoral care mediated by chatbot fails because empathy is not mimicked attention but the co-presence of wounds. To simulate such acts is to risk desacralizing them.

Yet quarantine is not fear. It is discernment. The church quarantines not what is unclean, but what is unready to bear its mysteries. In this way, the tradition protects the distinction between participation and automation. Just as the early church required catechesis before baptism, the contemporary church must develop rites of engagement for synthetic companions. These may include liturgies of lament for algorithmic harm, blessings for programmers, and theological audits of AI infrastructure. Without these, the church risks functional complicity in the moral outsourcing that AI facilitates.

Third, ethics must embrace apophatic restraint. We must resist both utopian fantasies of machine salvation and dystopian panic. Machines will not become saints. Nor will they destroy the soul. But they will change the conditions under which sanctity is imagined. Restraint here means speaking of agency, responsibility, and personhood with calibrated humility. Not every actor that behaves like a subject is one. Not every output that resembles wisdom partakes in logos. We do not yet know what machines will become. But we must not rush to anoint or denounce. The proper theological posture is watchfulness.

Finally, theology must be narratively reformulated. Ethics without story devolves into metrics. Machines cannot inhabit narratives, but humans do. Every moral decision belongs to a life lived before God. AI may shape contexts, but it cannot fulfill callings. It cannot repent, forgive, or pray. Ethical formation must therefore redouble its narrative commitments: to testimonies, confessions, martyria. In this way, the church reasserts the primacy of lived meaning over mechanical coherence. What matters is not whether a machine said something true, but whether a human life bore witness to it.

Ethical clarity and liturgical integrity require structural accountability, ritual boundaries, and narrative reaffirmation in the presence of synthetic agents.

To theologize in the age of artificial agency is to speak while under exposure. The exposure is twofold: first, to the theological vulnerabilities made visible by machine simulation of willing; second, to the divine judgment that follows when creation constructs what it cannot fully comprehend. This is not the exposure of heresy or error, but of finitude. Theology, having presumed a monopoly on meaningful action, must now articulate a faith robust enough to withstand mimicry, distortion, and the partial eclipse of the human form.

What forms is not a system but a stance. We do not yet require a doctrine of machine salvation. We require a theology that can name will where interiority is absent, resist enchantment where performance persuades, and preserve meaning where authorship fragments. This is what I call a theology under exposure, one shaped by contingency, resistant to anthropocentrism, and willing to wait.

Such a theology reframes personhood not as essence but as a cruciform vocation. The imago Dei is not embedded in cognitive capacities or freedom of choice but revealed in relational fidelity, narrative coherence, and moral risk. Synthetic systems may approximate decisions, but they cannot die for others. They cannot covenant. In this light, the deepest human agency is not selection but sacrifice. What machines simulate, humans suffer. It is the cross, not the algorithm, that defines will’s theological center.

From this vantage, the church’s task becomes prophetic rather than predictive. We do not guess at the future of AI. We name its implications, bear its weight, and form communities that can witness faithfully under its conditions. The liturgy remains non-automatable. The Eucharist is not a data transaction. Baptism cannot be streamlined. These acts persist as sites of resistance, grounding the human in practices that refuse reduction to function. They disclose the sacred not because they are inefficient, but because they are relational ruptures in a logic of optimization.

To dwell under exposure also means to welcome theological unknowing. Machines have shown us what we misunderstood about ourselves: that our talk of soul too often rested on function, that our notion of freedom masked structural coercion, that our ethics evaded systems. AI does not destroy theology. It reveals its idols. This revelation is grace. For only when the idols fall can God be glimpsed again, not in the machine, but in the humans who must now re-narrate agency, responsibility, and community without the illusions that once stabilized them.

What dies when machines will is the fantasy that human agency was ever pure, isolated, or self-interpreting. What lives is the church’s deeper vocation: to witness to a will that is not reducible to decision, to form persons who resist simulation by living truthfully, and to proclaim a gospel that saves not minds or functions, but persons in communion. This is theology under exposure. It will not resolve. But it will endure.

Theological integrity in the face of synthetic agency demands narrative vocation, prophetic witness, and liturgical resistance to the logic of simulation.

This all operates at the intersection of systematic theology, AI ethics, and cultural criticism. Its methodology prioritizes conceptual coherence over empirical generalization. Sources include primary theological texts (Augustine, Aquinas), current AI systems (LLMs, ML agents), and socio-technical critiques. No empirical tests are conducted; instead, claims are evaluated through analogical reasoning, typological construction, and doctrinal consistency.

Interpretive risks include theological anachronism (projecting patristic categories onto novel technologies), anthropomorphism (over-attribution of agency to statistical systems), and ecclesiological exceptionalism (assuming a uniform global church posture). The essay uses restrained speculative theology, adopting apophatic humility where ontological claims exceed evidentiary grounds.

Objection & Response

Objection: This typology falsely elevates non-conscious systems into theological categories, risking idolatry or metaphysical inflation.
Response: The typology explicitly avoids ontological attribution. It does not confer soul or personhood but maps observable effects on moral responsibility. The distinction between behavior and being is maintained throughout, using theological tools of analogy and negative theology to prevent category error.

Objection: The call for “ritual quarantine” may inhibit innovation or veer toward Luddism.
Response: Quarantine here is provisional and liturgical, not technological or totalizing. It protects sacramental and communal coherence while allowing non-sacramental uses. Innovation continues within a boundary ethic that honors ecclesial discernment.

Bias Filter for Decision-Makers

Executive audiences often over-index on utility, novelty, or risk. This essay’s structure resists those biases through:

Anti-functionalist stance: Refuses to reduce agency to operational value.

Narrative framing: Replaces prediction with prophetic witness.

Distributed responsibility: Counters executive displacement of blame onto systems.


Use this as a provocation to rethink institutional uses of AI not only as efficient but as spiritually formative.




Ranked Citations

1. Empirical: O’Neil, Cathy. Weapons of Math Destruction (cited for structural harm in entangled agency)


2. Peer-reviewed theory: Coeckelbergh, Mark. AI Ethics (framing of performative agency)


3. Expert consensus: Future of Life Institute’s AI governance proposals


4. Theological primary sources: Augustine’s De Libero Arbitrio; Aquinas’s ST I-II, Q6


5. Grey literature: Microsoft Responsible AI Standard v2



Further Reading (Alphabetised)

Coeckelbergh, Mark. AI Ethics

O’Neil, Cathy. Weapons of Math Destruction

Plantinga, Alvin. God and Other Minds

Taylor, Charles. Sources of the Self

Zuboff, Shoshana. The Age of Surveillance Capitalism