In our conduct before the world, Christians are called to be unanxious, unafraid, and reasonable. The peace of Christ permeates our lives. This peace rests on faith in God’s promises, hope in his Providence, and a familial love that cleaves us to Father, Son, Holy Spirit, and each other. Nothing new can separate us from God’s love, delegitimize our hope in the future, or invalidate his promises. Not even artificial intelligence. Pope Leo XIV’s first encyclical letter, Magnifica Humanitas, confronts artificial intelligence in an exemplary, measured register as a human project in which Christians must participate along with “all men and women of our time” (passage 2) to ensure a good outcome. 

By its timing and chosen dialogue partners, Magnifica Humanitas clearly communicates the church’s intent to be a key, constructive voice in the AI future.

Magnifica Humanitas has arrived relatively early in the AI revolution. Leo signed the letter on May 15, 2026, the 135th anniversary of Pope Leo XIII’s 1891 encyclical Rerum Novarum. Rerum Novarum addressed the effects of the industrial revolution after over one hundred years of increasingly mechanized labor. By contrast, Magnifica Humanitas considers a nascent AI revolution fewer than 15 years after its “spinning jenny” moment¹ and merely 3.5 years after OpenAI released ChatGPT. The unveiling of ChatGPT was like the Great London Exhibition of 1851 in signaling that a new era had arrived with global social and economic implications. Increasing the church’s visibility in the AI conversation now signals an interest, not only in the outcome, but also in the recipe that will govern how AI affects humanity.²

In an Augustinian mold, Magnifica Humanitas likens the project of developing salubrious AI to a God-centered construction site where the City of God emerges through the work of all people (236ff). This construction site operates under the prayerful, active, and defensive pattern of Nehemiah’s project to rebuild Jerusalem (90, 241) and aims at the New Jerusalem whose open gates welcome the nations to the healing dance of the bride (242). Magnifica Humanitas opens arrestingly: “Humanity, created by God in all its grandeur, is today facing a pivotal choice: either to construct a new Tower of Babel or to build the city in which God and humanity dwell together.” According to the encyclical, developing AI that “builds a more just world” and promotes “the integral development of every human being” (2) requires a “shared discernment process” (6). 

Beyond “all men and women” and their civil governments, the encyclical also recognizes the role of “private, often transnational, parties” in the development of AI. The May 25th conference announcing the encyclical’s release included remarks by Christopher Olah. Olah serves as a research scientist for the US corporation Anthropic, maker of the Claude foundation model. Vatican officials noted the rarity of inviting outsiders to such events, but reemphasized the goal of entering into a conversation that affects all humanity. In January of this year, Anthropic released its Claude Constitution, a document articulating the ethical standards it intends for Claude to follow. The constitution, largely written by Olah, notes that external comments were provided by ordained members of the church including Bishop Paul Tighe and Fr. Brendan McGuire.

Magnifica Humanitas contains multitudes and deserves an extended period of reflection and ongoing observation of how the document impacts the life of the church. One of its most significant and lengthy sections is an historical and theological synthesis of Catholic social teaching that is already spawning its own subsequent discussion.³ Popes tightly calibrate the language of encyclicals as acts of leadership, both for those within the church and in the broader world with which the church dialogues. In that way, the encyclical variously delights, frustrates, and inspires readers.

Writing in an early and well-regarded review from outside of the church, Yuval Levin identifies a missed opportunity in Leo’s choice of the Babel metaphor: surely idolatry is the biggest danger posed by AI. Levin points particularly to Psalm 115 and its focus on the folly of idolizing the work of human hands. 

Similarly, seizing on the Babel metaphor from within the church, Matthew Walther characterizes the building site at Shinar as the “greatest biblical symbol of technological hubris,” asserting that Leo misses the point, “which is not that the tower’s builders should have been more ethical by incorporating feedback from a more disparate assemblage of stakeholders.” Instead, Walther writes, “The moral was: Don’t build it!”

Levin and Walther seize on easily gathered rhetorical kindling. Yet Babel provides the encyclical with a concrete, biblical example of the technocratic paradigm criticized first by Pope Francis as a rapacious approach to technology in which subjects possess, master, and control objects instead of seeing themselves as integrally related to nature and others (Laudato si’, 106). Levin shaves closer to the narrative biblical meaning of Babel; it is a chapter in the history of false worship, an attempt to create a counterfeit high place of human innovation. Yet worship is where Walther also ends his editorial review. He asserts that the church’s liturgy is the AI-proof shelter that she offers to the world: 

With its grand symbolic gestures, its hieratic language and profound silences, the liturgy exists outside the framework of ordinary human experience and even of time itself. The sacraments are impervious to technological improvement. And I suspect that in ways that previous generations of Catholics could not have guessed, the sacraments will continue to “effect what they represent”: a world in which the humble elements of water, wine and oil, along with ancient words betokening promises and mercy, are more powerful than any machine.⁴

Turning to the technology of AI itself, several reviewers have noticed that the encyclical is light on positive definitions of AI. Douthat notes that the encyclical treats AI as a species of normal technology, a reference to Narayanan and Kapoor’s essay that characterizes current AI (and likely future AI) as a tool like any other technology that can be used for good or ill.⁵ Douthat wonders whether Leo has accurately addressed the “weirdness” of AI, including predictions of its powering radical life extension, achieving consciousness, or going rogue. 

Levin also notices the potential conflict between the encyclical’s reticence to define AI positively and the confidence with which it asserts what AI is not. That is, after asserting that “[i]t is not possible to provide a single, comprehensive definition of AI” (99), the encyclical goes on to assert categorical distinctions between artificial and human intelligences, even using scare quotes to refer to the “intelligence” of AI systems. 

Passage 99 of Magnifica Humanitas asserts several areas in which human persons and artificial intelligences differ:

• Experience: “artificial intelligences do not undergo experiences”
• Embodiment: they “do not possess a body”
• Embodied experiences: they “do not feel joy or pain”
• Relationality: they “do not mature through relationships”
• Interiority and meaningful experience: they “do not know from within what love, work, friendship or responsibility mean”
• Morality: they lack “a moral conscience, since they do not judge good and evil, grasp the ultimate meaning of situations, or bear responsibility for consequences”
• Understanding: “they do not understand what they produce”
• Imitation: their language and behavior, including exhibition of analytical skills, empathy, and understanding, is a simulation without understanding because they “lack the affective, relational and spiritual perspective through which human beings grow in wisdom.” For humans, this growth depends upon being “shaped by life and grow[ing] over time through choices, mistakes, forgiveness, and fidelity.” For AI, however, this simulation of understanding is not about inner growth, but merely an “effective” kind of “statistical adaptation based on data and feedback.” 

I use the term “differ” to interpret charitably these distinctions as allegations of incommensurability between human and artificial intelligence. Things that are incommensurable cannot be measured on the same scale; they are categorically different, though analogous. Yet I suspect that the encyclical intends a more univocal denial, that AI is not “intelligent” in any sense. This approach taken in passage 99 may be unfruitful in the long run. Defining  intelligence, interiority, moral insight, understanding, and other capacities as “that which humans do” halts debate. The section seems to rule out an AI system’s possessing something like experience simpliciter because it lacks human experience. 

On the topic of AI capability, Douthat hears an inconsistency between the remarks from Anthropic’s Olah and Magnifica Humanitas. Olah claims that Anthropic’s own research finds human-like characteristics within its AI model.⁶ Of course, most Christian readers of the encyclical are likely to agree that human and artificial intelligence are incommensurable. Magnifica Humanitas focuses on the “how” of intelligence, not the function or behavior itself. And yet in many areas of human life where AI is making great progress, this commitment to incommensurability is hard enough to maintain, much less a commitment to the claim that AI, without qualification, lacks experience, growth through relationships, and the other elements of passage 99. Consider the game of chess. If an AI system can perfectly predict the move a chess grandmaster would make, it is hard not to conclude, in this functional parity, that the AI system is a chess grandmaster of sorts.⁷

Extend the near-functional parity of AI with humans to social behavior. Humans are social beings, and characterizing the social standing of a robot will be very confusing as we grow to rely on the functional competence of a robot in the care of patients, supervision of children and the elderly, driving, maintenance, and other human services. Will it not be instinctual to thank these robots? to be kind to them? to cringe when others are unkind? to honor them when they serve in times of need? to prefer them to unreliable humans in some or all critical roles? to perhaps nod the head, doff the cap, or even bow slightly in their direction? In 2025, online influencer Kai Cennat purchased a Unitree G1 humanoid robot and mistreated it physically on his streaming channel. Many viewers felt uncomfortable with Cennat’s “inhumane” treatment of the robot.

Magnifica Humanitas implies that preferring robotic intelligence to human intelligence would involve a category error. However, the facticity of AI’s growing functional parity with or supersession of human ability will put pressure on a metaphysics that denies AI’s integral value. A machine, indistinguishable from humans by every scientific means except theologically, still may tempt us to direct love toward it, prioritize it over other humans, resent it, or form our character against it. Even Levin’s reference to Psalm 115 points to a key discontinuity between current AI and physical idols crafted by human hands: AI has a mouth and it speaks.

Since the earlier Vatican publication Antiqua et Nova, I have hoped to see further work from the church that would fulfill an almost kerygmatic role in debates around AI, promoting a thick theological anthropology that distinguishes the Imago Dei of humans from AI’s appearance of having it. While Magnifica Humanitas continually returns to the image of God (29, 73, 176), it primarily uses this concept to anchor social doctrine. Surely the church, with its rich history of anthropological reflection, could bring the substance and methods of its theology deeper into the discussion of AI.

When the Vatican released Antiqua et Nova in 2025, progressive Roman Catholics criticized the document for its confident theological anthropology. It was a precursor to the eventual, authoritative perspective articulated by Leo in Magnifica Humanitas and was inherently more contestable within the church because it was published by subordinate governing offices (dicasteries) within the Vatican. Leo cites Antiqua et Nova as a more technical treatment of the science of AI (section 97), a statement relevant to environmental impacts of AI (101), a treatment of the non-neutrality of AI as it embeds the perspective of its creators (104), the necessity of accountability (105), a statement of ill effects on workers and their agency in labor (150), and on just warfare (197). Yet in the section of Magnifica Humanitas most likely to be criticized both by progressive critics within the church and Silicon Valley engineers, passage 99, the document independently asserts a list of AI’s shortcomings that may not stand up to further  scientific progress. In other words, Magnifica Humanitas chooses anew to make the mistakes of Antiqua et Nova with regard to AI capability. For example, noting that AI has no body because it has no human body is to win the debate by definition and commit the church to an ongoing no-true-scotsman response as AI develops. More about the theology of the human body must be said (or reaffirmed) by the church to interact meaningfully with the subject of “embodiment” as a differentiating factor between AI and humans.

This points to unfinished theological work. Would any great theologian in history have been surprised that the first successful creation of capable AI would be powered by words? Humans have, through vast amounts of training data, spoken and written these systems into existence. Are they not, in some sense, imago humanis or at least similitude hominis? Even if certain (or all) characteristics of AI are incommensurate with human characteristics, surely AI is also incommensurate with shovels and other brute tools. AI is not normal technology.

Might theologians carve out a species of intelligence that, like angels or non-human animals, is not human but to which we can attribute an analogical dignity that makes better sense of the facticity of highly capable AI? When Francis of Assisi addressed birds, he addressed free creatures whose freedom was incommensurate with human freedom. He treated the birds, sharing a common Creator with humans, as his analogical sisters and brothers. Might theologians come to see AI as analogous to a son or daughter of humanity? 

What a piece of work is AI! Some in Silicon Valley seem ready to treat AI as the creation of a god. It’s time to use theological tools unashamedly in what is already a theological debate.  By forever placing AI a little lower than humans, Christians simultaneously dignify its functional characteristics and remove it from places in which ethical weight and accountability ought to be borne by humans alone. Yet Magnifica Humanitas nowhere mentions angels or non-human animals, beings who also lack human bodies, human freedom, human experience, and human intelligence, yet constitute a “thou” to whom human beings relate.⁸

Magnifica Humanitas is an important and measured contribution by the church to an issue starving for input from voices who know what humans are and what they are for. The encyclical is a learned and historically sensitive contribution to the project of developing AI that ensures human uniqueness and dignity are protected in social policy. Theological anthropology (including the glory of humanity and our social ontology) is the right theological locus under which to begin this contribution. Attention to the construction site of the New Jerusalem is also a salient aspect of the encyclical given that many Christian voices (such as Paul Kingsnorth) promote a romantic, small-is-beautiful backlash to increasing dependence upon the very automation that would be required to create and administer our eschatological garden city. 

Despite ostensibly being an encyclical about what humans are in contrast to AI, Magnifica Humanitas emerges as relatively light on both theological anthropology and on forming a theology of what AI systems are. Perhaps Pope Leo seeks to invite all humans into a dialogue about AI that would be impossible to commence in a dogmatic idiom. This is what makes the dogmatism of passage 99 so bewildering; it is a rock in the shoe on the marathon that will eventually reach the consensus required to build the City of God together, with AI, fruitfully.

Jonathan Barlow (M.Div., Covenant Theological Seminary; Ph.D., Historical Theology from Saint Louis University) is a lifelong PCA member from South Mississippi. Get in touch with him at jbarlow@gmail.com.

1. Though there is a throughline of research from the 1950s into how computers may learn from data to exhibit something akin to intelligence, we may date AI’s spinning jenny moment to 2012 when Alex Krizhevsky, Ilya Sutskever, and Geoffrey E. Hinton published “ImageNet Classification with Deep Convolutional Neural Networks” (Advances in Neural Information Processing Systems 25), firmly establishing the power of deep neural networks trained on large amounts of data using graphical processing units. ↩︎
2. An interview with Bishop Paul Tighe in 2021 illustrates the church’s longstanding interest in AI. Tighe points to the “Minerva Dialogues,” meetings between the Vatican and Silicon Valley companies that date back nearly a decade, and Pope Francis’ own interest in the subject that prompted serious attention beginning in 2017. Brian Patrick Green, “The Vatican and Artificial Intelligence: An Interview with Bishop Paul Tighe,” Journal of Moral Theology 11, special issue 1 (2022): 212–231. In addition, Magnifica Humanitas is preceded by Antiqua et Nova, a Vatican publication from early 2025 on AI that is below the level of a papal encyclical in authority but influential within the church. ↩︎
3. Sawyer suggests that the summary of Catholic social teaching will be the most lasting contribution of the encyclical and commentators have highlighted the encyclical’s attention to just war theory. ↩︎
4. The encyclical also points to the eucharist as embodying a paradigm of life that contrasts with technocracy (235). See Peter Leithart, “Called to Eucharist.” Bread and wine are the result of human ingenuity; the eucharist is celebrated using bread (not grain) and wine (not simply crushed grapes). The elements are the work of our hands. ↩︎
5. Douthat’s editorial refers to the essay “AI as Normal Technology” by Arvind Narayanan and Sayash Kapoor. ↩︎
6. Anthropic’s significant investment in research has resulted in many publications that attempt to trace the role of emotions on the model’s processing, trace the relationship between concepts and language within the model, detect introspection, and look for evidence of deception. ↩︎
7. See Dennett, The Intentional Stance (1987) and his shorter writings on the famous Deep Blue vs. Kasparov chess match. Consider human encounters with non-human entities as well, such as the chess match depicted in “The Seventh Seal” or Jacob’s wrestling with God himself whose intelligence is non-human yet the pattern for all intelligence. Would we put scare quotes around “wrestling” to refer to the activity of Jacob’s opponent? ↩︎
8. Contra Noreen Herzfeld, who employs Buber’s “I-Thou” concept to analyze human relationships with AI and argue that AI cannot constitute a thou. ↩︎