Babel Without Jerusalem
Pope Leo XIV’s encyclical on artificial intelligence opens with a biblical choice: raise a new Tower of Babel or rebuild Jerusalem with the patience of Nehemiah. The choice is sound; the development, defensive. The problem comes afterward: almost the entire text is devoted to warning against Babel and very little to explaining how Jerusalem gets rebuilt. The first readings, mostly admiring, have overlooked this imbalance.
That retreat hides a twofold error. First, within the Church itself: the encyclical declines to explore how AI might restore an organized, specialized Catholic mission of moral intermediation, today a weak one. Second, in economics: when it speaks of work, it still thinks as if dignified employment could be protected by decree, without bothering to understand how it is produced.
The Pope is right to note that AI is not neutral. Every technical system encodes priorities: what it measures, what it ignores, what it optimizes, what it rewards, what it punishes. A moral AI would be no innocent calculator. It can reinforce biases and psychological comforts. It can grant us absolution with perfect courtesy. It can replace the effort of self-examination with the calm of a costless emotional absolution. It can make easy what ought to remain hard.
The line between a useful moral tool and an atrophying oracle depends on exactly that: AI serves when it forces us to think, not when it thinks for us.
Beyond this, the encyclical stops. And it stops at a revealing point. When it turns to conscience, it regards all mediation with suspicion. In §51 it cites John Paul II approvingly: “the respect owed to the journey of conscience” is “a positive achievement of modern culture.” It is not Protestant by doctrine, but its moral architecture shows Protestantizing reflexes: it treats conscience as a space to be shielded from all intermediation, not as a faculty that also needs formation, contrast, judgment and expert help. Across 245 paragraphs on human dignity before the machine, confession, spiritual direction, casuistry and the training of confessors never once appear. A Catholic pope who, faced with AI, reasons as if ecclesial mediation were part of the problem and not part of the solution. This is no accident: Leo XIV is an Augustinian, and it was that vein—the primacy of interiority over mediation—that centuries later fed the Reformation.
For centuries, Catholicism built a dense moral technology based on confession, the examination of conscience, spiritual direction, the analysis of intention, circumstance, gravity, habit, reparation and even casuistry, with prosocial consequences. That effort produced abuses, as every human work does. But it also produced a specialized form of applied moral reasoning and proved essential to the development of the kind of individual that defines Western civilization.
The encyclical says much about dignity, truth, freedom, education. It says little about that institutional function proper to and definitive of the Church: helping to form the moral judgment of concrete people in concrete cases. And that is precisely where AI could be valuable: to assist discernment, train the clergy, order cases, compare moral traditions, pose demanding questions and reduce the arbitrariness of pastoral counsel.
The Pope diagnoses well and prescribes badly. In §107 he writes: “A more moral AI would be of no use if that morality is decided by a few.” But the answer he gives is political. The classic Catholic answer would have been another: to bring to that discussion an organized tradition of moral reasoning that few institutions in the world still preserve.
The opportunity is greater because the scarcity is also economic. Personal services of spiritual accompaniment and formation can hardly scale without losing quality. They grow relatively more expensive as industry raises its productivity. The Church has fewer vocations and fewer trained priests. AI does not solve that scarcity, but it can ease it. It is surprising that an encyclical on AI does not dare to raise the point.
The second error appears when it speaks of work. The encyclical rightly sees that employment is not only income: it is identity, responsibility, participation. It is also right to warn against an automation that surveils the worker or turns him into an appendage of the machine. But its economic diagnosis is incomplete. It treats employment as a good to be preserved, without explaining how it is created. In §163 it puts it plainly: “in the age of AI and robotics, it is no longer possible to rely solely on the market’s ‘invisible hand.’” There it gives itself away: it has decided that the economy is a problem of moral orientation, not of incentives.
Dignified employment is not decreed. It is produced: with productivity, investment, human capital, flexible organization, competition and technological adaptation. If automation is blocked or loaded with badly designed obligations, the firm reacts. It substitutes capital for labor when labor grows costly. It passes costs through to prices. It hires less. It outsources. It automates at another point in the chain. Or it disappears.
Protecting the worker is not the same as shielding his job. The opposite: today’s shielded job is tomorrow’s lost employment, the job of the young worker who never gets in and of the veteran who becomes surplus. Labor dignity demands more than good intentions: it demands understanding the incentives that sustain employment.
The same error of underrating production runs through both planes. In morality, it is not enough to invoke dignity and accompaniment; moral judgment must be produced, and that demands institutions, specialization and experience. In economics, it is not enough to invoke dignified work; sustainable employment must be produced, and that demands productivity, human capital, investment and a market.
Conscience and employment have something in common: both degrade when we try to protect them without understanding how they are produced.
The encyclical is effective as a warning and weak as a program. It knows how to say “beware.” It struggles to say “let us build.” It even cites Tolkien to urge us to “do what lies in our hands,” but it does not understand—and therefore does not explain—what, concretely, lies in the hands of the Church.
It fears that AI will replace man. And rightly so. But it should also fear the opposite: that, for fear of replacing him, it will decline to help him. For centuries, Catholicism understood that human freedom is defended not by leaving it alone but by forming it. That tradition does not need a machine that absolves. It needs instruments that compel us to think, examine, repent and improve.
The Pope himself says it: Babel or Jerusalem. But Jerusalem is not raised by abandoning the work. It is raised with good blueprints and better builders. AI must be in those blueprints to empower the builders. The encyclical forgets that the good is not preached. It is built.
Translated with AI from the Spanish version published at The Objective, 31 May 2026.