Where a papal encyclical on artificial intelligence and secular humanism converge — and where, beneath the shared demands, they cannot both be right.

An analytical essay by Bob Reuter, President of AHA Lëtzebuerg

Read soberly, with the incense cleared away, Pope Leo XIV's Magnifica Humanitas — his first encyclical, that is, a formal teaching letter addressed to the whole Church — is startling for how much of it a secular humanist can simply endorse. And we can now be precise about what the humanist position is, because the global humanist movement has written it down in two recent texts. The first is the Vienna Manifesto on Digital Humanism (Vienna, 2019), drawn up by the computer scientist Hannes Werthner with a circle of scientists and scholars, calling for technologies to be shaped in accordance with human values rather than the other way round. The second is the Luxembourg Declaration on Artificial Intelligence and Human Values, adopted in July 2025 by the General Assembly of Humanists International (the worldwide federation of humanist organisations) and built around ten ethical principles for the development and governance of AI.

Set the encyclical beside these two texts and the practical overlap is almost embarrassingly close. The Luxembourg Declaration's first principle holds that AI must never displace human judgment and that decisions deeply affecting people's lives must remain in human hands; the Vienna Manifesto likewise insists that choices touching human rights continue to be made by humans — which is exactly the encyclical's worry about surrendering morally weighty decisions to opaque, automated systems. The Declaration's principles of the common good and of democratic governance — that AI should serve humanity rather than enrich a privileged few, and that no corporation or nation should wield unaccountable power — restate the encyclical's alarm at technological power concentrating in private, border-crossing actors beyond democratic control. The critical wing of digital humanism presses the point further than Rome does: in Digital Humanism (2022) the critical theorist Christian Fuchs reads that concentration not as an accident to be corrected but as the logic of digital capitalism itself, so that what the encyclical diagnoses as a moral disorder he diagnoses as a structural one, rooted in who owns the means of computation. Its principle of transparency and autonomy, and the Vienna Manifesto's demand for the transparency and accountability of algorithms, match the encyclical's call for contestability. Protection from harm, shared prosperity, the defence of human creators, the integrity of truth against disinformation, and responsibility to future generations and the environment — each of these Luxembourg Declaration principles has a near-twin in the encyclical, down to its anxiety about the manipulation of what it calls the "collective imagination" (the shared stock of images and stories through which a society pictures reality). On the level of demands, the encyclical could almost be a religiously phrased restatement of the two humanist texts. This is no accident, since all three draw on one European inheritance: the Enlightenment, the philosophy of Immanuel Kant, the human-rights tradition, and the early-twentieth-century movement called personalism, which placed the irreplaceable worth of the individual person at the centre of ethics.

The overlap reaches deeper still, to the nature of the machine itself. The encyclical denies that AI genuinely understands anything; so, on wholly secular grounds, do the leading digital humanists. In Digitaler Humanismus (2018) Julian Nida-Rümelin argues, by way of Searle, that a computer only shuffles signs to which we — not it — assign meaning, so that understanding, judgment, and responsibility remain ineliminably human. Luciano Floridi adds the complementary point that AI's real novelty is to have decoupled successful action from any need for intelligence: the machine acts without grasping. The conclusion is Rome's; not a word of it depends on the soul. That this route reaches Rome's conclusion does not make it settled, however: Searle's Chinese Room is disputed among secular philosophers themselves. Daniel Dennett — who coined the term intuition pump in reply to exactly this argument — holds that it persuades by the vividness of its staging, the lone clerk shuffling slips of paper, rather than by argument, and that it quietly sets aside the systematic complexity in which understanding, if it is anything, would have to consist. On that view the flat denial that a machine could understand is less a finding than a picture, and the question stays open.

Such disagreements run within the secular camp; they do not disturb the broader point — the sheer breadth of agreement between the encyclical and humanist thinking, on the policy demands and, for most, on the nature of the machine alike. That agreement has not gone unremarked within the humanist community itself. The Austrian humanist Andreas Gradert, among others, has read Magnifica Humanitas against the Vienna Manifesto and concluded that, once its theological frame is set aside, the encyclical largely restates positions long established in digital humanism and secular AI ethics — the real difference, on his account, lying at the level of ultimate justification rather than in the demands themselves. That the overlap exists, and that humanists have remarked on it, is by now familiar ground.

The purpose of this essay is to go beyond it. Granting the convergence in full, it asks whether the disagreement truly stays confined to the choice of foundation — and argues that it does not. The differing grounds are not merely two warrants bolted beneath one shared ethic; they yield different first-order commitments — on the worth of autonomy, the meaning of human limits, and the ends of life — that no amount of agreement on AI policy can dissolve. Once the theology is set aside, substantive disagreements remain, and it is these that the rest of this essay sets out to map.

THE FOUR DIFFERENCES

The divergence begins at the foundations themselves. The humanist documents are unusually explicit about theirs, and it is precisely there that they part from Rome. The Luxembourg Declaration roots its principles in "reason, evidence, and our shared humanity", and closes by naming its values as reason, compassion, dignity, and freedom. The Vienna Manifesto appeals to the critical rational reasoning of the Vienna Circle — the celebrated interwar group of scientifically minded philosophers in Vienna — and treats reason, freedom, and responsibility as a single chain: we are free because we can reason, and responsible because we are free. In both, the ground is the human being's own reason and shared humanity. In the encyclical, the ground is God. And the gap that opens there does not reduce to the mere word "God". Four differences are essential, and each would survive the removal of theological ornaments.

I - The Source of Dignity

The first concerns where human dignity comes from. By "dignity" all three texts mean the basic, unearnable worth that forbids treating a person as a mere instrument. The encyclical grounds that worth in the person's existence as a creature willed and loved by God: it is built in from the start, independent of anything one can do or achieve, and present in full even where reason or consciousness is absent. The humanist documents ground worth instead in something about the being itself — its capacity to reason and to govern itself, and its capacity to have interests and to suffer. (Philosophers call this the question of "moral status": what it is about a being that makes it matter morally in the first place.) The two foundations agree on a great deal — including, in the Declaration's sixth principle, that technological progress must never erode human dignity by reducing people to their output, which is also the encyclical's protest against valuing persons by efficiency. But they come apart at the edges of life. From dignity-built-into-existence the encyclical derives an absolute right to life "from conception to natural end" and condemns both abortion and assisted dying; from worth grounded in autonomy and interests, mainstream humanism reaches the opposite verdict. The Luxembourg Declaration is a document about AI and does not itself legislate bioethics, but it is the same value-base — reason, autonomy, and the interests of the actual person — from which humanists defend bodily self-determination, and it is exactly there that the two groundings visibly diverge.

II - The Value of Autonomy

The second and deepest difference is the value placed on autonomy itself. Here the two traditions use nearly the same word and mean opposite things by it. The Luxembourg Declaration makes autonomy a named principle — individuals deciding how their own data is used, and able to query, contest, and shape the systems around them — and its culminating principle, "Human freedom, human flourishing", sets out to "maximise human freedom". The Vienna Manifesto wants technology bent to human will rather than the reverse. For the humanist, in short, the expansion of human self-determination is the very point. The encyclical's governing image runs the other way. It is drawn from two biblical stories: the people of Babel, who build a tower to heaven to "make a name for ourselves" and are scattered for their pride, and Nehemiah, who rebuilds the ruined walls of Jerusalem as humble, cooperative work with God at the centre. Within that frame, the wish to be the "sole author of oneself" is the sin of Babel. Strip the theism away and a value-judgment still stands: that the drive to self-determination is a danger to be tempered by humility, not a good to be maximised. This is the reverse of what Kant called Mündigkeit, the maturity of daring to think for oneself instead of deferring to authority, and of what the freethought tradition calls libre examen, the weighing of every claim by one's own reason rather than on someone's say-so. The quarrel is not about whether God exists; it is about whether the self-authoring human being is an achievement or a tower-builder courting collapse.

III - The Meaning of Human Limits

The third difference concerns the meaning of human limits. The Luxembourg Declaration's picture of flourishing is frankly ambitious about overcoming them: it hopes to turn AI toward advances in science and medicine, toward resolving global challenges, and toward freeing people for leisure, learning, and richer connection. Limitation, on this view, is largely something to push back against. The encyclical takes the opposite stance: vulnerability, fallibility, and mortality are not defects to be engineered away but the very soil in which compassion and love take root, and the genuine way to become "more than human" is through grace — a gift from God that raises us beyond our nature — never through technological self-improvement. A humanist will object that, taken at full strength, this would condemn anaesthesia and vaccines as readily as it condemns the dream of the upgraded human, and that praising limitation comes more easily from comfort than from a sickbed. Yet this is where the humanist should tread carefully, because the strongest form of the encyclical's point survives the loss of God. The German sociologist Hartmut Rosa argues, under the heading of Unverfügbarkeit (roughly, "uncontrollability"), that the modern compulsion to make everything controllable and available on demand hollows out our experience — a critique that needs no theology at all. The real disagreement is therefore narrower than it first looks: not whether the urge to control everything can be self-defeating, but where to set the limit, and whether to call that urge a "sin" or merely a disorder of modern life.

IV - Foundations and Method

The fourth difference is at once about foundations and about method, and it lies beneath the other three. The humanist documents make it unusually clear. They locate authority in reason and evidence — public, shareable, and open in principle to revision by anyone who can argue better. The encyclical locates it in continuity with tradition: a teaching is to be trusted because it stands in line with earlier popes and councils, the body of official Church teaching known as the "magisterium". That is a system able to correct itself from within but not genuinely open to being overturned by an argument from outside it. The encyclical adds that human reason, left to itself, is ultimately not enough — that without fixed foundations anchored finally in God, human rights have nothing solid to stand on and decay into a contest of power, in which the strongest decide what counts as a right. The humanist wager, written into both declarations, is the contrary: that an adequate ethics can be built from human materials alone — reason, shared interests, mutual agreement. One should not caricature the difference as the charge that the encyclical is "unfalsifiable" — falsifiability, the philosopher Karl Popper's test that a scientific claim must be refutable by some possible observation, is not a fair demand to make of any ethics, religious or secular. The precise difference is that the humanist texts are argument-open where the encyclical is argument-closed.

IN SUM

What emerges, then, is not a modest religious surcharge added onto a shared secular core, but two opposite structures of thought that happen to meet at the roof. The encyclical and the two declarations can co-sign almost the same list of demands — human oversight of consequential decisions, democratic control of AI, transparency, protection of the vulnerable, the priority of the common good — while disagreeing profoundly about the human being those demands are meant to protect. They read the same digital landscape from opposite directions. One orients the human being upward, toward an end given to us from beyond ourselves; the others hold that the human being is the author of its own ends. In the vocabulary of philosophy, the first is a vision of heteronomy, of being ruled from outside or above, and the second of autonomy, of being ruled from within — and that divide governs how each grounds not only morality but knowledge itself. Precisely where the two directions genuinely diverge — over the ends of life, over the worth of self-determination, and over whether human limitation is to be overcome or honoured — the conclusions diverge with them. The deepest difference, in the end, is not a belief that humanism happens to lack. It is a direction. The overlap is the shared map of the terrain. The difference is which way is up.