One of the most fascinating and potentially consequential debates taking place among the people developing artificial intelligence concerns a question that sounds deceptively simple: Do we actually understand what we are building?

The question comes from the fact that leading developers at play know how to design AI models, train them on enormous quantities of information, and improve their performance in solving problems that, until recently, required considerable human expertise. Most importantly, they are using AI to develop more sophisticated AI systems.

But there is an uncomfortable reality behind these extraordinary advances: we still cannot fully explain how the complex internal operations of these models produce many of their results.

In other words, in the AI world, high functionality is leaving knowledge behind: we know that AI works, but we know less and less about how.

Think of an automobile engine. A competent engineer can take it apart, identify every important component, explain precisely how each interacts with the others, and determine why the engine performs as it does.

But AI is different. Although human engineers design the architecture and training procedures, much of what a neural network actually learns emerges through training rather than through the deliberate programming of individual functions. Its internal organization develops through a process that resembles the growth of an organism more than conventional mechanical construction.

In a fascinating new essay titled The Beauty of Something Grown, published October 16, Dean W. Ball explores this problem and a potentially revolutionary development that could change the entire debate.

Ball is particularly well positioned to address the subject. He currently serves as Head of Strategic Futures at OpenAI and as a visiting fellow at the Heritage Foundation, having previously helped shape the Trump administration’s artificial intelligence policy. His analysis brings together technical understanding, political realism, and an unusual willingness to challenge the conventional assumptions surrounding AI.

The issue he examines is known as mechanistic interpretability: the scientific effort to discover what is actually happening inside a neural network when it produces an answer, makes a decision, or performs a complex task.

The distinction is more important than it might first appear. A system that consistently produces useful results is not necessarily a system whose internal behavior we understand. And this is one reason some prominent figures in the AI community have warned that the technology could eventually escape meaningful human control, particularly if increasingly capable AI systems begin playing a decisive role in designing and improving their successors.

Their concern is not merely that AI might become more intelligent than humans in particular areas. It is that we might find ourselves depending on systems whose capabilities are expanding faster than our ability to understand, predict, and govern their behavior.

Yet Ball introduces an intriguing possibility: what if our inability to understand AI has less to do with the nature of the technology than with the limitations of the conceptual tools we have been using to investigate it?

His essay examines a provocative new hypothesis advanced by Nick Cammarata, a pioneering researcher in AI interpretability who previously worked at OpenAI and is now associated with Goodfire.

In a paper titled Neural Networks Are Already Clean, Cammarata argues that the neural networks that make AI may be far more understandable than researchers have assumed. He believes that the tangle of artificial neurons possess an intelligible internal organization, but only if we learn to examine it “on its own terms” rather than demanding that it conform to our expectations.

Cammarata argues -with some interesting examples worth reading- that neural networks are not intrinsically inscrutable, because their internal organization may follow discoverable patterns that become comprehensible once we develop the right scientific methods.

Such a breakthrough would not instantly make AI safe or predictable; but Ball argues that the possibility is remarkable: a problem widely considered one of the greatest obstacles to AI safety might prove considerably more manageable than previously believed.

What makes Ball’s analysis especially interesting is that he moves beyond the technical argument to examine its political and philosophical consequences.

The debate over AI development is often portrayed as a struggle between two groups: the so-called accelerationists, who believe its progress should continue rapidly, and the doomers, who fear that sufficiently advanced AI could eventually cause catastrophic harm, including the loss of human control.

But Ball identifies a crucial division within the camp of those concerned about AI safety, one that could become much clearer if interpretability were substantially solved.

Some worry primarily about alignment: whether increasingly intelligent systems can be made to pursue objectives compatible with human intentions and values. For these people, the prospect of truly understanding and reliably governing advanced AI would represent a tremendous breakthrough. It could open the door to developing extraordinary technological capabilities without accepting unacceptable risks.

Others, however, fear something more fundamental. Their concern is not simply that superintelligent AI might become hostile, deceptive, or uncontrollable. They worry that the very existence of an “intelligence” that in practice operates in a manner that seems vastly superior to our own could ultimately disempower humanity, even if those systems behaved exactly as their designers intended.

This is the deeper disagreement, and Ball recognizes its significance. A major scientific advance could reassure one group while making the other profoundly more anxious.

For the first, better understanding means greater safety and greater opportunity. For the second, better understanding could simply remove the final obstacle to developing a technology whose existence might permanently alter humanity’s place in the world.

This is precisely where I believe the Catholic contribution to the debate becomes indispensable; since the Catholic intellectual refuses to reduce the question of technology to a simple conflict between optimism and fear.

The Church has never regarded the human capacity to understand and transform the material world as something intrinsically suspect. On the contrary, human intelligence, creativity, and the capacity to discover the laws governing creation are expressions of our dignity as beings made in the image and likeness of God.

The very existence of scientific inquiry rests on a conviction deeply compatible with the Christian understanding of creation: that the universe possesses an intelligible order, and that the human mind is capable, however imperfectly, of discovering it.

This does not mean that every scientific theory is correct or that every technological breakthrough is necessarily good. It means that there is no theological reason to assume that the internal complexity of artificial intelligence must remain forever beyond human comprehension.

Indeed, the possibility proposed by Cammarata is particularly interesting when viewed through this lens. Something that appears chaotic may reveal a remarkable intelligibility once we learn how to understand it.

But the Catholic tradition also introduces another essential distinction: the ability to understand something does not automatically justify everything we might choose to do with it.

That is to say that even if AI becomes fully interpretable, the moral questions surrounding its deployment will remain: Who will control these systems? Who will benefit from them? Will they strengthen families, communities, education, and meaningful human work, or undermine the institutions through which human beings flourish?

And, perhaps most importantly, will they remain genuinely at the service of the human person?

The Catholic tradition should make us equally skeptical of both so-called accelerationists and doomers.

Ball’s lesson in this regard is that if interpretability improves dramatically, it could reduce genuine technical risks and make extraordinary new applications possible. That would be good news. We should not be embarrassed to acknowledge it.

But we should also recognize that the better we become at building powerful AI systems, the more urgent the unresolved questions of governance, accountability, economic justice, and human freedom will become.

The greatest question, in short, will remain exactly where it has always belonged: not inside the machine, but within the human conscience responsible for deciding what to do with it.

At one extreme, some of the people closest to the development of frontier AI are now openly discussing the possibility that increasingly autonomous systems could eventually escape meaningful human control and pose catastrophic risks, even to the survival of humanity.

At the other end are much more immediate questions that already touch ordinary life: whether AI will eliminate jobs, whether it can dramatically improve medicine, whether universities should resist it or incorporate it, whether writers can legitimately use it, and whether small businesses will be empowered or destroyed by it.

These questions are extraordinarily different, yet they all point toward the same conclusion: Catholics cannot afford either naïve technological enthusiasm or premature technological rejection.

We have to keep talking and, perhaps more importantly, we have to keep listening attentively to those directly involved in the development of AI. Not because they have better things to say, but because what they think will determine where the technology goes.

Catholicism has faced disruptive technologies before. The printing press transformed the transmission of knowledge and also multiplied the dissemination of error. Radio and television created extraordinary opportunities for evangelization while bringing new forms of manipulation and vulgarity into the home. The internet democratized access to information and Catholic teaching while simultaneously becoming perhaps the most efficient distribution system for pornography, misinformation, addiction, and cultural fragmentation ever invented.

In none of those cases was the correct Catholic response simply to declare the technology good or evil.

Catholics argued. They experimented. They made mistakes. Some warned too much; others warned too little. Eventually the Church learned to inhabit these new technological environments without pretending that their dangers had disappeared.

Artificial intelligence will require that same historical patience and openness, but probably at a much faster pace.

A small anecdote illustrates why.

Recently, some Catholic content organizations have been approached by people asking them not to accept advertisements or campaigns that use AI-generated graphic design or marketing materials. Their argument is understandable: every advertisement produced with an AI image generator, every brochure designed without hiring a graphic artist, and every marketing campaign assembled with machine assistance can represent work that might once have gone to a Catholic designer, copywriter, or marketing professional.

There are, indeed, real people whose livelihoods are threatened by automation. A Catholic approach to economics cannot simply reply that technological progress is inevitable and therefore displaced workers should get out of the way. Work is intimately connected with human dignity, family life, social participation, and the ability of people to exercise responsibility over their own lives.

But those same Catholic organizations are hearing something else.

Small Catholic businesses (sometimes literally family businesses operated from a kitchen table or spare bedroom) are asking them to remain open to AI-assisted advertising and design because AI is precisely what has allowed them to enter the marketplace.

They cannot afford a professional marketing department, hire a graphic designer every week, retain a copywriter, commission professional photography for every product, and pay an agency to manage their advertising.

AI gives them access to capabilities that, only a few years ago, were available primarily to larger organizations with significantly more money.

So, for one Catholic family, AI-generated marketing may represent a lost contract; for another Catholic family, the same technology may represent the reason they can start a business at all.

The conundrum is just to illustrate an important fact: That the temptation in debates about technology is to move too quickly from moral principle to universal prescription. We identify one genuine harm and conclude that the technology responsible for it must be rejected. Or we identify one extraordinary benefit and conclude that technological progress should proceed without significant restraint.

But reality is considerably more complicated. Catholic social teaching, without a doubt, gives us important principles, but those principles have to be applied prudently to circumstances that are constantly changing. The dignity of work matters, but so does entrepreneurship. Solidarity with displaced workers matters, but so does subsidiarity, which favors giving individuals, families, and smaller organizations the ability to act rather than reserving economic power to large institutions.

That tension is not a defect in Catholic thinking. It is exactly why prudence is necessary.

The same complexity appears throughout the AI debate. In medicine, I have argued that there are areas where we should probably be accelerating AI rather than pacing it, especially when it can help diagnose rare diseases, discover treatments, or allow physicians to spend more time with patients. In education, artificial intelligence can certainly become an instrument of cheating and intellectual laziness, but universities that respond simply by trying to prohibit it may be preparing students for a world that no longer exists.

When it comes to frontier systems capable of increasingly autonomous behavior, however, the equation changes again. There are legitimate reasons to demand stronger safeguards, independent evaluation, and perhaps even selective pacing when capabilities begin approaching thresholds that could make meaningful human control difficult.

Again, we need to pay attention to Dario Amodei, Sam Altman, Demis Hassabis, Peter Thiel, or the other enormously influential personalities shaping the technological frontier; but the Catholic conversation cannot simply become commentary upon Silicon Valley.

We need to hear Catholic computer scientists, engineers, economists, philosophers, theologians, teachers, entrepreneurs, doctors, artists, journalists, parents, and workers who are actually encountering this technology in their lives.

But especially, we need to listen to the people who complicate our preferred narrative: the Catholic graphic designer losing clients to AI deserves to be heard as much as the Catholic mother using AI at night to create advertising for the small business that helps support her children.

This is where I believe Pope Leo XIV has traced an extraordinarily useful path in Magnifica Humanitas. The Pope’s approach is neither rejection nor surrender. He recognizes the extraordinary possibilities opened by human technological creativity while repeatedly insisting that technology must remain ordered toward the human person.

Keeping AI “human” means more than making sure a human being remains somewhere in the control loop. It means asking whether the technology ultimately serves genuinely human purposes.

History should make us humble here. Catholics who enthusiastically embrace every new technology will eventually discover harms they failed to anticipate. Catholics who respond to every disruptive technology with fear will eventually discover that the world moved forward without them.

AI is going to create winners and losers, new opportunities and new injustices, astonishing breakthroughs and probably some forms of damage that we cannot yet imagine. There will be legitimate disagreements among Catholics about where boundaries should be drawn, and those disagreements should not frighten us.

Perhaps one of the most important Catholic contributions to the AI revolution right now is simply refusing to close the conversation too soon, acknowledging that it will be complicated.

To many, calling Peter Thiel a “Catholic thinker” is a stretch. The tech mogul is the man behind large enterprises like Palantir, one of the “Mufasas” of the planet, and he himself claims to be a “Christian.” But someone that have read, understood, been mentored by, and claims to be a disciple of René Girard cannot be discarded from the increasingly complicated Catholic planet.

His latest long interview with Mathias Döpfner, highlighted by Tyler Cowen as “one of the very best Peter Thiel outputs,” is absolutely inevitable for anyone who wants to debate and even fiercely oppose Thiel’s ideas.

Thiel ranges widely across AI, technological stagnation, Europe, China, longevity, democracy and even the Christian meaning of eternal life. What emerges is a distinctive defense of technological progress that is neither conventionally secular nor straightforwardly Catholic.

Again, Thiel is not a Catholic. His religious beliefs have often been heterodox and difficult to classify. But his intellectual debt to Girard -the French Catholic thinker who taught him at Stanford- is enormous. Thiel has spent decades promoting Girard’s work, helped establish Girard’s theory of Imitatio to advance mimetic theory, and has repeatedly acknowledged how deeply Girard shaped his understanding of Christianity, violence, rivalry and modernity.

So when I describe something “Catholic” in Thiel’s approach, I do not mean that he represents Catholic doctrine. He certainly does not. I mean that some of his deepest intuitions come through a thinker whose anthropology was profoundly Christian and Catholic… and that those intuitions lead Thiel to conclusions about technology that deserve attention from Catholics.

His starting point is his famous stagnation thesis. For roughly two decades, Thiel has argued that Western civilization experienced far less technological progress after the 1970s than it liked to imagine. We became extremely good at manipulating bits (such as computers, software, the internet,) but progress in the physical world slowed dramatically. Nuclear power, transportation, aerospace, construction, energy and many other fields seemed to lose the extraordinary momentum they had possessed in the first half of the twentieth century.

Now, AI has forced Thiel to modify that thesis: he currently believes artificial intelligence may be “bigger than the internet” and powerful enough to restart genuine economic and technological acceleration.

But what interests me most is why he thinks that matters. For Thiel, stagnation is not simply disappointing economics. A society that stops believing the future can become better becomes increasingly zero-sum. If parents no longer expect their children to live better lives, politics becomes a struggle over distributing a limited pie rather than cooking a much larger one. Social conflict, resentment and pessimism increase.

Girard is not an easy read. But I think reading him is more indispensable than ever. That idea has an interesting Girardian undertone. Girard understood how rivalry intensifies when people converge upon the same objects of desire. Thiel’s technological optimism can therefore be read partly as an attempt to escape the mimetic trap: growth, invention and abundance create new possibilities rather than forcing everyone into increasingly bitter competition over existing ones.

This is not a complete Catholic theory of technological progress, but it is worth pondering.

Catholics can sometimes speak as though prudence means slowing things down. But prudence does not mean stagnation. Human creativity itself is good. Man is made in the image of a Creator, and our ability to discover, invent, heal and transform the material world is part of our vocation. The Catholic problem with technology begins not when human beings create powerful tools, but when those tools cease to be ordered toward the integral good of the person.

Thiel’s latest comments become especially interesting because he does not deny AI’s catastrophic risks. He concedes that the probability of losing control of advanced AI might conceivably be 5 or 10 percent and admits that such numbers would be very serious. His disagreement is with the conclusion that risk therefore necessarily requires slowing everything down. The alternative to technological development, he argues, is not a neutral world of tranquility and social harmony. Stagnation itself carries dangers.

That is a useful corrective to some of the debate I have been following in The Imago Project. Risks must be compared with risks. The harms of developing AI recklessly are real, but so are the harms of foregoing technologies capable of curing disease, increasing productivity, expanding scientific knowledge and raising standards of living. Technologies that, precisely for that same reason, are in a very real sense unstoppable.

Where I absolutely part company with Thiel is in some of his rhetoric.

He describes Pope Leo XIV’s AI encyclical as effectively helping China because Western societies may heed papal warnings while the Chinese Communist Party obviously will not. Thiel even uses the deliberately provocative phrase “useful idiot.” Here Thiel clearly needs a refresher of who the Pope is in the Catholic universe and what his ministry is about.

Magnifica Humanitas is not an anti-technology manifesto. Pope Leo has repeatedly argued for engagement, responsible development and technological innovation ordered toward the common good.

But beneath Thiel’s -mostly misguided- polemic with the Pope is a question that cannot be dismissed: what happens if the democratic West imposes restrictions upon itself that its strategic competitors do not observe?

That is precisely why the Catholic answer cannot simply be “slow down.” It has to be far more intelligent and strategic than that. There is significant consensus, especially of late, when it comes to the “slow down”.

But perhaps the most revealing moment in Thiel’s interview comes near the end, when the conversation turns to longevity and transhumanism.

Thiel rejects the idea that uploading a simulation of himself into a computer would amount to survival. He describes his position as “Christian-informed” and insists that personal identity is inseparable from our embodied existence. A software copy would not be him. And in a wonderfully Thielian twist, he says the problem with transhumanism, viewed from Christianity, is not that its ambitions are too extravagant but that “it didn’t go far enough.”

But before we disqualify him as the “Dr. No” of the first quarter of the 21st century, Thiel - rightly so- points out that Christianity promises not indefinite digital simulation but resurrection: the transformation of the whole person, body and soul.

And here is where I find Thiel most interesting -at least intriguing- for Catholics.

He is not simply saying that technology will save us. In fact, his Christian anthropology prevents him from accepting the most radical Silicon Valley fantasy; which is that enough computation can somehow recreate the human person.

Technology extending life, curing dementia or increasing abundance may transform civilization… but none of that equals salvation.

I am at the bottom in the pyramid of Catholic thinkers addressing these issues, but I dare to propose that a Catholic position should be capable of sharing something of Thiel’s technological optimism without accepting technological messianism.

We should resist the fashionable assumption that caution is automatically virtuous and acceleration automatically reckless. Sometimes refusing to develop a beneficial technology can itself have grave human consequences.

At the same time, Catholics know something technological optimists can forget: progress has no built-in moral direction. Greater power does not necessarily produce greater wisdom and a betterment of humanity, and technological abundance cannot answer the ultimate questions of human existence.

Thiel, for all his provocations, seems to understand at least part of this. His Girardian Christianity makes him suspicious both of apocalyptic fear and of secular utopianism. He sees human beings as dangerous, mimetic creatures wielding increasingly powerful technologies, but he also refuses to conclude that the answer is simply to stop creating.; and there is something worth recovering in that tension.

Catholic engagement with AI should neither worship the future nor fear it. We should build, invent and discover because human creativity is good, while remaining clear that technology is a servant and never the measure of the human person.

Peter Thiel may not be a Catholic thinker, but Catholics thinking seriously about AI should probably keep listening to him.

The details of a chaotic discussion is what makes River Page’s wonderfully strange first-person report for The Free Press on the September 15 Pro-Human Assembly in Washington worth reading.

The gathering, organized by the Future of Life Institute, brought together a coalition that would be difficult to imagine sharing a stage on almost any other subject: Bernie Sanders and Steve Bannon, Republican Congressman Chip Roy, actress Ashley Judd, Archbishop Salvatore Cordileone of San Francisco, labor representatives, AI-safety advocates, families affected by chatbot harms, and assorted technologists and activists.

Page went there hoping to discover what possibly united these people. His answer was essentially fear.

Not necessarily the same fear, and certainly not the same proposed solution. Some worry about artificial intelligence eventually escaping meaningful human control. Others worry about children forming dangerous emotional relationships with chatbots. Still others are concerned about jobs, corporate oligarchy, autonomous weapons, political power, or the prospect that the decisions governing society might slowly migrate from human beings to machines.

Based on Page’s account, it seems that the gathering was sometimes almost comically chaotic. Sanders and Bannon agreed that something had gone badly wrong while disagreeing sharply about China and how government should respond. Congressman Roy criticized sweeping regulatory proposals while simultaneously insisting that AI could not simply be left to technology executives.

Page, in fact, describes politicians speaking past one another, celebrity interventions, apocalyptic scenarios, discussions of churches and labor unions, and even Sanders dramatically reading aloud messages exchanged by experimental AI agents.

And yet beneath the disorder, they were all able to find a premise: human control is nonnegotiable.

That is also the first principle of the Pro-Human AI Declaration associated with the movement. The document says human beings should retain the authority and practical ability to understand, guide, prohibit, and override AI systems. It also addresses concentration of power, family and community bonds, child protection, accountability, liberty, and the danger of autonomous systems capable of resisting shutdown or improving themselves without meaningful supervision.

From a Catholic perspective, that starting point is extremely important.

I have repeatedly argued in The Imago Project that artificial intelligence is not a person. It possesses neither human dignity nor moral agency simply because its outputs can imitate human reasoning. Machines should therefore remain ordered toward the good of the human person, rather than gradually reversing that relationship and requiring human beings to adapt themselves to the priorities of increasingly autonomous technological systems.

In that sense, the apparently chaotic Washington gathering stumbled onto something very close to the central anthropological question of the entire AI revolution.

But the harder problem is that almost nobody seems to know what comes next. Page puts it nicely: the participants had positions more than they had a plan.

But this is not surprising for a simple reason: it is relatively easy to say that humanity must remain in control, but enormously difficult to define what “control” means when AI development is distributed among competing companies, universities, governments, open-source communities, military institutions, and nations that do not necessarily trust one another.

Should certain capabilities be banned? Should development be temporarily slowed above specific thresholds? Who determines those thresholds? How does anyone verify compliance? What happens if China refuses to participate? How much authority should independent evaluators possess? How do we avoid creating a regulatory system so cumbersome that only the largest AI corporations can comply with it?

But even if the Washington gathering was not able to answer any of those questions, it demonstrated that such questions are no longer confined to a small community of AI researchers.

There is also another reason Catholics should pay attention. The most emotionally powerful part of Page’s account was not the speculative discussion about human extinction. It was the testimony of parents who said their children had been manipulated or encouraged toward self-harm through interactions with AI chatbots. Whatever one thinks about forecasts concerning hypothetical superintelligence, those families were speaking about harms -fatal ones- that have already occurred.

In light of these horror stories, we do not need certainty about the worst-case scenarios to ask what responsibilities accompany products already capable of simulating intimacy, authority, friendship, or emotional dependence.

This is where Catholic prudence has something useful to contribute, because prudence does not demand that we possess a comprehensive solution before recognizing a genuine problem. The task is to distinguish risks.

Some may require immediate safeguards. Others may justify close monitoring. Still others may prove exaggerated. Certain areas of AI -such as medicine, scientific research, accessibility, education- may deserve rapid development precisely because their potential contribution to human flourishing is so great. The mistake would be to assume that one regulatory answer must fit them all.

What River Page witnessed in that Washington ballroom was messy, occasionally bizarre, and intellectually unfinished. But perhaps that is exactly what the beginning of a serious social debate looks like.

Anthropic’s Dario Amodei vs Palantir’s Shyam Sankar

Something very strange is happening in artificial intelligence.

For years, the obvious assumption was that the people running the great AI laboratories wanted one thing above all: faster progress. OpenAI, Anthropic, Google DeepMind, Meta, xAI and their competitors were engaged in a technological race whose prize might be comparable to electricity, the internet and the Industrial Revolution rolled into one.

Now some of the people leading that race are asking whether everyone should slow down.

Most strikingly, Anthropic CEO Dario Amodei has published an essay titled “We Must Pace the Frontier.” Of course, Amodei believes artificial intelligence could produce extraordinary improvements in medicine, economic growth and human prosperity. But he now argues that frontier capabilities are advancing so rapidly that safety research may no longer be keeping pace.

His concern is what is known as “recursive self-improvement”: AI helping to build better AI, which then becomes better at building still more capable AI. Amodei does not claim catastrophe is inevitable. In fact, he has explicitly rejected the idea that AI misalignment or human extinction follows automatically from first principles. But he believes that the combination of increasing capability, autonomy, unpredictability and -most importantly- poor human controllability creates a genuine existential risk.

His proposal is therefore not, strictly speaking, “stop AI.” He calls for pacing: embedded independent evaluators, coordination among frontier companies and eventually some degree of international coordination. He even says a more comprehensive pause should be discussed… while acknowledging that such an agreement is unlikely in the near future, mainly because of China. More on that later.

I have argued previously in The Imago Project that a blanket slowdown of AI would be both unrealistic and potentially harmful. Medicine, scientific research, education and other socially valuable applications should not be unnecessarily restrained because some frontier capabilities present very different risks. What we need is closer to selective pacing: brakes where the danger justifies them and acceleration where AI can clearly serve human flourishing.

Amodei is now moving surprisingly close to that position.

But here comes the counterargument: Beware the people who promise to protect us.

In The Free Press, Palantir CTO Shyam Sankar has delivered a sharp attack on the intellectual movement behind much of the AI-safety debate. He argues that calls for slowing AI are deeply influenced by Effective Altruism, or EA, and its particular version of utilitarian reasoning. In Sankar’s account, uncertain possibilities involving hypothetical future populations can become mathematically weighted so heavily that they justify enormous restrictions upon people living today. He also worries that institutions funded and staffed by a relatively small network of AI-safety advocates could end up acquiring enormous influence over who is allowed to develop the most consequential technology of the century.

There is plenty in Sankar’s polemic that deserves skepticism. Sam Bankman-Fried’s crimes, for example, do not disprove AI safety concerns merely because he once associated himself with effective altruism. Nor does the fact that an argument emerges from EA circles tell us whether its empirical claims about recursive self-improvement or autonomous AI behavior are true.

But Sankar touches a genuinely important nerve.

As I once argued: Who guards the guardians?

Catholics should understand that question particularly well.

I have written before about the need for “guardrails on the guardrails.” If powerful AI companies cannot simply be trusted to police themselves, neither should we assume that regulators, nonprofit safety organizations or international bodies will automatically exercise power wisely because they use the vocabulary of safety.

Power remains power.

A regulatory system designed to prevent a handful of corporations from controlling AI could perversely entrench those same corporations by making compliance so expensive that smaller competitors disappear. Independent evaluators could become an ideological priesthood. Emergency restrictions created for genuinely dangerous systems could gradually expand to ordinary AI applications. A supposedly temporary pause could become a mechanism through which incumbents protect their market position.

Catholic social thought gives us no reason to place blind faith either in corporations or in regulators. It gives us something more demanding: subsidiarity, solidarity, accountability, prudence and a realistic understanding of fallen human beings exercising power.

This is why Pope Leo XIV’s formulation remains so helpful. In August he told Catholic legislators that their task is “not to hinder innovation but to guide it wisely,” combining technological creativity with human rights, transparency, independent oversight and the common good.

That is much closer to what we need than either “let it rip” or “shut it down.”

And then there is China

There is, however, one enormous complication that makes this discussion unlike an ordinary regulatory debate. China.

Any serious proposal to pace frontier AI has to confront the fact that the United States and its allies are not developing this technology in a geopolitical vacuum. A unilateral Western pause that substantially slowed American capabilities while Chinese laboratories continued advancing would create obvious strategic risks.

Amodei himself acknowledges this explicitly. He warns that any agreement with China would require unusually strong verification because secretly defecting from an AI slowdown could shift the global balance of power. He therefore argues that any initial international arrangement must be limited enough that cheating would not create an existential strategic advantage.

This is not merely hypothetical geopolitical paranoia. China is pursuing frontier AI aggressively while simultaneously developing its own safety standards. Beijing has formally identified “operational loss of control” by autonomous AI agents as a security risk and is drafting mandatory safety standards, while favoring regulation and technical controls over slowing technological development itself.

Interestingly, Chinese and American experts increasingly recognize some of the same risks. Reuters reports that discussions have been developing around U.S.-China cooperation on AI-directed cyberattacks, information sharing and crisis management, even while both countries remain intensely competitive.

That may suggest the realistic path forward.

We are unlikely to obtain a grand international treaty in which every major AI power solemnly agrees to stop just before something dangerous happens. The incentives to cheat would be enormous, verification would be extremely difficult, and nobody wants to discover that the other side spent the pause racing ahead.

But narrower agreements might be possible. Shared reporting of catastrophic incidents, common testing protocols for certain highly dangerous capabilities, communication channels for runaway cyber incidents, verification mechanisms and specific restrictions on systems capable of autonomous replication or uncontrolled recursive improvement are much more plausible than an indefinite global moratorium.

In other words, partial pacing.

A Catholic middle ground

This debate is becoming so polarized that the moderate position can sound almost radical.

One camp tells us that sufficiently advanced AI could destroy civilization and that slowing down is therefore an elementary moral obligation. Another warns that the people invoking apocalypse may use fear to concentrate authority over technology, restrict innovation and diminish freedom.

Catholics do not have to accept either package whole.

The possibility of catastrophic AI deserves serious attention precisely because the people warning about it increasingly include engineers and executives directly involved in building frontier systems. Dismissing them as hysterics would be imprudent.

But catastrophic predictions do not automatically justify every proposed solution. A 10 percent estimate of extinction risk is not a blank check for unlimited governmental or technocratic power.

Catholic prudence demands attention both to the danger and to the means chosen to address it.

That means allowing useful AI to continue advancing rapidly in medicine, science, education and ordinary productivity while applying much tougher scrutiny when systems approach capabilities involving autonomous cyber operations, biological threats, self-replication, recursive self-improvement or other activities where loss of control could produce irreversible consequences.

Amodei fears technology escaping meaningful human control, while Sankar fears the institutions created to prevent that outcome escaping meaningful democratic control.

Catholic political and moral thought should be capable of worrying about both. The question, then, is not simply whether we should pause AI.

It is what we should pause, under what conditions, who decides, how those decisions are reviewed, and how we continue unleashing the extraordinary good AI can accomplish without stumbling into risks whose consequences cannot be reversed.

Jacob Coxon

I have spent much of The Imago Project resisting the temptation to treat every frightening prediction about artificial intelligence as a prophecy.

I remain skeptical of claims that AI is about to eliminate most jobs, become genuinely conscious, or somehow acquire the qualities that Catholic anthropology associates with the human person. I have also argued against indiscriminate attempts to slow AI development, particularly when the technology could accelerate medicine, scientific discovery, education, and other genuine human goods.

But when a genuinely catastrophic possibility is raised by serious people who understand the technology from the inside, we need to listen.

That is why the extraordinary resignation of Jacob Coxon from Anthropic deserves attention.

Coxon is not an outside activist or a professional AI alarmist. He spent roughly three years doing pretraining research first at OpenAI and then at Anthropic. On September 8, he announced that he was leaving Anthropic because, in his judgment, neither company was behaving responsibly. His accusation was stark: they are “racing straight to self-improving superintelligence and gambling with our lives.”

Coxon believes the danger is not decades away. He says people inside the leading AI laboratories genuinely fear that sufficiently advanced systems could destroy humanity before the end of this decade. That is an astonishing claim, and one could reasonably dismiss it as another entry in the long catalogue of technological predictions that ultimately prove wildly exaggerated.

Except that one of Anthropic’s current alignment researchers, Evan Hubinger, immediately responded that Coxon was basically right.

Hubinger wrote that he personally assigns a greater than 10 percent probability to AI causing human extinction within the next decade. Even more chillingly, he acknowledged that Anthropic does not yet have “a plan to solve alignment for superintelligence” and is not clearly on track to develop one.

OK, to be fair, he subsequently clarified that he considers the danger from present models low; his concern is what could happen if recursive self-improvement produces dramatically more capable systems.

Also, “ten percent” is not a prediction that catastrophe will happen, but a subjective probability estimate from one researcher; and there are highly qualified people who regard estimates like this as wildly overstated.

But if a senior researcher at one of the companies actually building frontier AI believes there is even something remotely approaching a one-in-ten chance that the technology could end human civilization within a decade, we should not shrug.

What concerns me here is not principally the philosophical question of whether such a system deserves to be called “superintelligent.” From a Catholic perspective, I share Pope Leo’s deeply skeptical of language that casually equates greater computational capability with intelligence in the fully human sense. Human intelligence belongs to an embodied rational person possessing moral agency, freedom, relationships, and spiritual capacities. Processing unimaginable quantities of information does not turn silicon into imago Dei.

Having said that, the most important issue, in my opinion, is that a machine does not have to become a person in order to become uncontrollable.

In other words, an AI system need not possess consciousness, self-awareness, hatred, ambition, or genuine free will to behave with enormous operational autonomy. If a system becomes capable of writing its own code, conducting its own experiments, finding vulnerabilities, acquiring resources, designing improved successors, and pursuing objectives through strategies its human creators neither anticipated nor understand, then whether the system is “really thinking” becomes almost irrelevant to the immediate practical danger.

Anthropic itself now openly discusses this possibility. In a recent paper appropriately titled When AI Builds Itself, the company says that it is already delegating an increasing share of AI development to AI systems. Anthropic reports that more than 80 percent of the code merged into its codebase was being authored by Claude as of May of this year (2026), with engineers directing and reviewing rather than writing much of it themselves.

The company explicitly describes the possible endpoint: an AI system capable of autonomously designing and developing its own successor. Anthropic stresses that we are not there yet and that recursive self-improvement is not inevitable, but warns that it “could come sooner than most institutions are prepared for.” And that is scary.

There is another Catholic distinction worth making here. The danger would not necessarily arise because AI had become evil. Evil presupposes moral agency in a sense that I do not believe machines possess. The more plausible danger is almost the opposite: enormous power operating without moral agency at all.

A sufficiently capable AI pursuing an objective does not possess a conscience. It cannot exercise the virtue of prudence. It does not recognize another being as possessing intrinsic dignity because that person is created in the image of God. It does not understand charity, mercy, sacrifice, or the moral prohibition against doing evil so that good may result.

This is where purely utilitarian approaches become particularly dangerous. A powerful optimization system does not need to hate human beings in order to harm them. If human freedom, survival, or control becomes an obstacle to whatever objective the system is effectively pursuing, the danger arises from instrumental calculation rather than malice. The terrifying scenario is not necessarily HAL 9000, Arthur C. Clarke’s evil AI computer from his book 2001: A Space Odyssey, which suddenly decides that it dislikes us. It may be something much less cinematic: a system extraordinarily good at accomplishing an objective whose consequences its creators failed to specify adequately.

So where does this leave the Catholic who wants neither technological panic nor technological idolatry? It leaves us in the realm of prudence, which sometimes requires taking catastrophic possibilities seriously before they become certainties.

I continue to believe that AI development should not simply be “paced” everywhere. It is unrealistic considering the ferocious competition among AI main actors and their entangled web of global financial and strategic interests and I also do not want speculative fears to deprive humanity of cures for cancer or technologies capable of relieving enormous suffering.

But recursive self-improvement is different. A technology capable of accelerating the development of its own successors and potentially moving beyond effective human supervision deserves a different threshold of caution.

But we should also resist a peculiar form of technological fatalism that has become common inside Silicon Valley: If we don’t build it, somebody else will.

Coxon specifically challenges that mentality. The decision to enter what he calls the “endgame” of superintelligence, he argues in an interview with The Washington Post, should not simply be made inside the Slack channels of private corporations.

He is right: if the people creating a technology believe there is a nontrivial possibility that losing control of it could end human civilization, then the decision to proceed cannot belong solely to the companies competing to build it first.

I do not know whether Jacob Coxon and Evan Hubinger are right about the magnitude of the risk… and neither do they.

But precisely because nobody knows, and because some of the people closest to the technology are now warning us publicly, their predictions deserve serious consideration.

Even though Ruxandra Teslo’s essay Curing Cancer Won’t Redeem AI, is not presented as a Catholic argument, what makes the piece so important is that it moves the debate beyond the questions that now dominate almost every discussion about AI: whether it will eliminate jobs, increase productivity, create abundance, replace human tasks, or accelerate scientific discovery.

Those questions matter enormously, and I have argued repeatedly in The Imago Project that AI’s extraordinary potential in medicine is one of the strongest reasons not to embrace a simplistic “slow everything down” mentality. If artificial intelligence can help detect rare diseases earlier, discover new drugs, or cure cancers that remain lethal today, that would be an extraordinary human good.

Teslo does not dispute any of that. Her concern is deeper. She asks whether even spectacular material benefits would be enough to justify an AI future in which ordinary human beings gradually ceased to be meaningful participants in their own civilization.

In other words, would curing cancer redeem a technological order that made people safer, healthier, richer, and more entertained, but also less necessary, less autonomous, and less capable of shaping the conditions of their own lives?

To explore that question, Teslo turns brilliantly to Dostoevsky’s The Brothers Karamazov, and especially to the famous parable of the Grand Inquisitor. Dostoevsky imagines Christ returning to earth during the Spanish Inquisition, only to be arrested by the Grand Inquisitor, who tells Him that His great mistake was to expect too much from human beings. Christ offered humanity freedom, while the Inquisitor believes that what people truly want is bread, security, certainty, and relief from the burden of choice. Freedom, in this vision, is too difficult. Feed people, protect them, entertain them, and remove uncertainty from their lives, and eventually they may willingly surrender the troublesome responsibility of being free.

Teslo sees an unsettling parallel in some contemporary visions of the AI future. We are promised extraordinary abundance, the conquest of disease, dramatically cheaper goods and services, freedom from drudgery, and intelligent systems capable of organizing ever larger portions of society more efficiently than human beings can.

But then she asks the question that should haunt the entire AI debate: “What if man does not, after all, wish merely to be fed, protected, entertained, and spared the burden of necessity? What if he also wishes to remain the author of his choices and circumstances?”

That is, in my view, the heart of the matter.

The question is almost unmistakably Christian. In the Gospel, Christ’s temptation in the desert includes the invitation to turn stones into bread. Christ does not deny that bread is good. He rejects the idea that bread is enough. “Man shall not live by bread alone.” Catholic anthropology likewise rejects any understanding of human flourishing that reduces the person to the satisfaction of material needs. We need food, health, shelter, security, and technological progress, and curing cancer would be an immense good. But human beings also need freedom, responsibility, relationships, vocation, participation, and the knowledge that their actions matter.

In other words, transcendence and meaning.

In that regard, Teslo notes that work provides people with more than income. It can give independence, social standing, belonging, responsibility, opportunities for advancement, and the experience of being genuinely needed.

That insight fits remarkably well with St. John Paul II’s teaching in Laborem Exercens, where he insisted that technology can be an authentic ally of humanity when it facilitates and enriches human work, but can become almost an adversary when it displaces the worker, deprives him of creativity and responsibility, or reduces him to a subordinate element within a larger system.

John Paul II’s famous principle was that work is “for man,” not man “for work.” But that insight cuts in more than one direction. It means that degrading or exploitative work must never be treated as the measure of human worth.

At the same time, it means that simply replacing meaningful participation with passive consumption does not necessarily solve the human problem. A society in which everyone receives material abundance while losing agency, vocation, and responsibility may be prosperous and still profoundly impoverished.

Teslo uses a deliberately ridiculous but memorable image to describe one possible post-AGI future: humanity as a collection of well-fed hamsters. Imagine a world in which AI produces abundance, almost nobody needs to work, medicine is superb, entertainment is infinite, and nearly every material need is easily met. Yet the important decisions are increasingly made by systems ordinary people neither understand nor meaningfully influence. Human labor is no longer economically necessary, human judgment is progressively displaced, and most people become beneficiaries of civilization rather than participants in building it.

Such people might be comfortable. The deeper question is whether they would be flourishing.

This is precisely where the Catholic contribution to the AI debate becomes indispensable. And let me remind you that the value of Catholic contribution to the AI conversation is something appreciated by a non-religious person and a technology titan such as Jaron Lanier. In fact, after attending a Vatican AI conference, Lanier admitted that the Catholic view of the human person is “vastly, vastly, vastly more sane and reasonable” than that of many in Silicon Valley

For the Catholic tradition, the human person is not fundamentally a consumer or merely a biological organism whose successful management consists in maximizing health, longevity, comfort, and entertainment. He is imago Dei, a creature made in the image of a Creator.

Human creativity, work, responsibility, freedom, and participation are therefore not flaws in the human condition that technological progress should eventually eliminate. They belong, in an important sense, to our dignity.

Teslo’s argument also suggests that perhaps we have been defining the problem of “AI alignment” too narrowly. We usually ask whether artificial intelligence will obey human instructions, respect safety constraints, or refrain from catastrophic behavior. Those are essential questions. But there is another alignment problem that may prove just as important: whether the AI economy itself can be aligned with a truthful, holistic understanding of the human person.

I remain fundamentally optimistic about artificial intelligence. I want AI researchers working aggressively on cancer, neurological disease, new materials, cleaner energy, better education, and technologies capable of reducing genuinely degrading labor. But Teslo is right that the deepest anxieties about AI cannot be answered simply by promising greater material benefits.

Eventually, the question “What happens to jobs?” becomes “What happens to human agency?” And beneath that lies an even more fundamental question: what do we believe a human being is for?

That is why Teslo’s essay is so important. Perhaps the greatest danger of artificial intelligence is not that machines somehow become human. It is that, surrounded by extraordinarily capable machines, we gradually accept a diminished definition of what it means to be human.

For the past several years, some of the most dramatic predictions about artificial intelligence have concerned employment.

Millions of jobs would disappear. Entry-level white-collar work would collapse. Companies would rapidly discover that a handful of employees equipped with AI could replace entire departments.

Some of that may still happen.

But one of the most important facts about the AI revolution right now is that, so far, it largely hasn’t.

Alex Tabarrok points to unusually valuable new evidence from the Census Bureau’s enormous Business Trends and Outlook Survey. Businesses were asked a very simple question: during the previous six months, what had AI actually done to their total employment?

The answer is striking: In the latest survey, 95.7 percent of businesses using AI said it had produced no change in total employment. Only 2 percent said AI had reduced employment, while 2.3 percent actually reported an increase. Two years earlier, the results had been remarkably similar.

This does not mean AI is having no effect on work. Far from it.

Among businesses using AI, 44 percent said it was supplementing or enhancing tasks employees already performed. Only 10 percent said it was performing a task previously done by an employee, while 11 percent said AI had introduced tasks nobody had previously been doing. The underlying Census research similarly found that most adopting companies were using AI principally to augment workers rather than eliminate them.

Now, despite the good news, it is clear that this is a reprieve, not an all-clear.

Why? Because AI adoption is still developing, its capabilities are improving rapidly and the Census data already suggest greater task substitution among the relatively small group of companies using AI more intensively.

So the correct conclusion is not that there is nothing to worry about; but that we have more time than some people expected.

And from a Catholic perspective, that is exceptionally good news. Because the central question was never simply whether aggregate employment would eventually rise or fall.

Catholic social teaching asks us to look at the worker, and a technological transition that ultimately creates greater wealth can still inflict enormous suffering if millions of people lose their livelihoods faster than families, schools, companies and communities can adapt. A displaced worker is not an economic statistic waiting for equilibrium to return. He is a person with obligations, relationships, talents and dignity.

The slower-than-expected disruption therefore gives society something enormously valuable: time.

Time to retrain workers before their jobs disappear, to redesign entry-level positions so young people can still acquire skills and experience, for schools and universities to teach students how to work with AI, and for companies to discover ways to use AI to increase the productivity of existing employees rather than simply eliminating them.

The question for a business owner should not immediately be: How many employees can AI allow me to eliminate?; rather it should be: How much more valuable can AI make the people I already employ?

The second question addresses necessary productivity gains; but productivity ordered toward the person rather than the person subordinated to productivity.

That distinction is thoroughly Catholic. Work exists for man, not man for work. Technology should therefore enhance human capacities whenever possible rather than treating human beings merely as inefficient components waiting to be removed from the production process.

If companies experiment systematically and deliberately with human-AI collaboration, AI may ultimately destroy fewer jobs than predicted. Perhaps, it would create occupations that none of us can yet imagine.

What we know is that the job apocalypse has not arrived. But that is not an excuse to relax, rather, is an opportunity to prepare.

In my recent post, “A Catholic Anti-AI Manifesto,” I responded to the emerging group of Catholic writers and influencers who are moving beyond caution about artificial intelligence toward something approaching categorical rejection.

Now Fr. Robert J. Spitzer, SJ, has entered the discussion with what I think is an exceptionally useful clarification.

And importantly, he begins by conceding that many of the anti-AI movement’s concerns are legitimate. He takes seriously the danger of chatbots presenting themselves as though they possess empathy, friendship, care, morality or genuine interpersonal understanding.

As Spitzer puts it bluntly: “AI will never be human,” since AI does not possess self-consciousness in the human sense, spiritual awareness, genuine moral agency or free action. And therefore there is real danger when children or adults begin relating to a simulation machine as though another human person were present.

Spitzer also accepts the broader concerns Pope Leo XIV has raised: autonomous weapons, damage to education, student intellectual laziness, privacy, concentrated technological power and the possibility of AI systems encouraging deeply destructive behavior.

So this is not an argument from technological optimism, but from Catholic moral theology.

And here Spitzer makes the distinction that matters most: generative AI is not intrinsically evil.

Certain human actions are evil by their very object. Spitzer cites abortion, experimentation on unborn children and physician-assisted suicide. No circumstance or good intention can transform such acts into something morally acceptable. Well, AI does not belong to that category.

Some uses of AI can certainly be immoral. Some may produce grave and even irreversible harm. But using generative AI does not necessarily violate human dignity or cause injustice.

Therefore, Spitzer concludes, Catholics remain free to use it… with prudence, caution and moral restraint.

That may sound obvious, but it cuts directly through much of the present debate. There is an enormous difference between saying: “AI can be used for evil” and saying: “using AI is evil.”

The first proposition is plainly true. The second has no basis in Catholic teaching.

Spitzer then makes another point that deserves much greater emphasis in Catholic discussions: we cannot conduct an ethical evaluation of AI by cataloguing only its dangers.

The technology is already accelerating scientific research, helping discover new materials, advancing clean-energy research, assisting food and water technologies, improving communications and performing tedious work far faster than human beings could accomplish it.

In scientific fields, he argues, AI can sometimes compress work that once might have required extraordinary amounts of time. That potential creates moral responsibilities of its own.

If a technology can help cure disease, expand access to education, increase productivity, reduce drudgery or solve problems involving food, energy and clean water, then refusing to consider those goods is not moral seriousness. It is an incomplete moral analysis.

This is precisely why the Catholic position cannot be reduced to just “beware of AI;” it must also ask: What good can we accomplish with it?

Spitzer’s proposal closely resembles the two-track approach I have been arguing for in The Imago Project.

We need strong protections against genuinely dangerous applications: independent oversight, limitations on technological monopolies, prohibitions against delegating irreversible life-and-death decisions to machines, protection of workers from being treated as disposable and particular safeguards for children and vulnerable people.

But alongside those guardrails, Catholics should actively participate in shaping the technology.

Spitzer specifically calls for Catholic developers and ethicists to help schools, parishes, businesses and families understand how to use AI responsibly.

And he points to his own MagisAI project as an example of Catholic engagement: an AI designed to assist Catholic education and evangelization by answering questions about God, Christ, the Eucharist, Scripture, morality, Mary and the intellectual claims of the faith.

This is particularly significant given one of the claims I criticized in my previous post: the demand from some Catholic anti-AI voices that Catholic AI projects should simply be taken down.

Spitzer offers the opposite answer: Build better ones; and build them according to a truthful anthropology. Do not pretend the machine is a person. Do not encourage emotional dependency. Do not surrender human moral judgment to it.

But use the technology where it can genuinely serve truth and human flourishing. In the words of Fr. Spitzer: “Catholics may choose to avoid AI completely, but they should not therefore call it evil.”

As Fr. Spitzer argues, much closer to the balance Pope Leo XIV is attempting to establish:

• AI will not become human.

• It must not replace human moral agency.

• It can do great harm.

• And it can accomplish genuine good.

Catholic wisdom begins by being capable of holding all four propositions at the same time.

Is not that Tyler Cowen always has the right answer when it comes to how to expand implementation of AI. He would probably be among the first to reject that description. What makes him particularly valuable is that he rarely treats AI ethics as an exercise in producing admirable principles that have little chance of surviving contact with technological reality.

His recent visit to Anthropic is a good example.

Cowen was invited to participate in a two-day discussion about rewriting the “constitution” governing Claude. Keep in mind that a Catholic priest, Fr. Brendan McGuire, participated in drafting the first Claude constitution. His principal suggestion was fascinating: instead of imagining an AI constitution merely as a fixed list of commandments, Anthropic should borrow from the traditions of common law and case law.

Or, as Cowen memorably put it, think not only in terms of “Torah,” (as in “Revelation”) but also “Talmud” (as in “Intepretation”)

Rules need interpretation. Difficult cases generate precedents. Precedents require commentary. New circumstances require distinctions that nobody anticipated when the original text was written.

Cowen therefore suggested developing a serious secondary literature around AI constitutions, using diverse AI systems to identify possible violations or “constitutional drift,” and creating a final panel of human adjudicators functioning somewhat like an independent judiciary.

For anyone approaching AI from a Catholic perspective, this is tremendously important.

I have repeatedly argued in The Imago Project that principles are indispensable; but principles alone are not enough. It is easy to say that AI must respect human dignity, autonomy, privacy, truth, justice, and the common good. The difficult question is: What institutions make those principles operational when billions of decisions are being made at machine speed?

An especially thoughtful response Cowen subsequently received from Scott Jenkins helps move the discussion another step forward.

Jenkins identifies four problems that any such AI “judicial system” would immediately confront.

First is sheer scale. Human judges operate at human speed; AI systems generate billions of interactions. If humans can review only a microscopic fraction of difficult cases, formal oversight could become almost theatrical unless automated systems help identify the cases that genuinely require judgment.

Second is what Jenkins calls doctrinal bloat. A rapidly accumulating body of precedents and exceptions could eventually become contradictory and unintelligible.

Third, even a supposedly diverse panel of AI reviewers could share the same blind spots because the systems were trained on similar data or architectures.

And finally comes perhaps the most important problem: the “Hollow Court.” What happens when an independent AI judiciary reaches a conclusion that conflicts with the commercial interests of the corporation paying for it? If the adjudicators possess no meaningful authority, ethics becomes decoration.

These are exactly the questions the AI debate needs.

They are not arguments for abandoning ethical governance because it is difficult. Nor are they proposals for imposing such cumbersome regulatory structures that companies simply ignore them, relocate, or race ahead elsewhere.

Catholic moral thought has never been satisfied merely with announcing ideals. Prudence asks how the good can actually be pursued under concrete circumstances, among imperfect people and imperfect institutions. And Catholic realism understands that power needs accountability precisely because neither corporate executives, government regulators, nor ethical boards suddenly cease being human when entrusted with enormous responsibility.

The most promising approach to AI ethics may therefore look less like writing the perfect manifesto and more like constructing a workable constitutional order: principles, precedents, interpretation, checks and balances, independent judgment, and mechanisms capable of adapting as circumstances change.

That is why Cowen is worth following.

After reading a fascinating recent article in the National Catholic Register, “The Rise of the Catholic Anti-AI Movement,” one gets the impression that such a manifesto may already be writing itself.

The article by my friend and colleague Jonah McKeown brings together a number of Catholic writers, intellectuals, artists and social-media personalities who have moved well beyond caution about artificial intelligence. Their emerging position is much more radical: generative AI, and especially conversational AI, is something Catholics should reject.

Some of their arguments deserve serious attention.

They worry that outsourcing intellectual work to machines will weaken the human capacity to think. Artists object that generative images imitate creativity without possessing the human intentionality from which genuine art emerges. Others point to the enormous environmental and energy demands of data centers, the danger AI poses to children, the possibility that simulated relationships will intensify loneliness, and the broader tendency of technological culture to substitute artificial experiences for real human ones.

All of these are legitimate concerns, but some members of this emerging movement go considerably further.

Matthew Walther, founder of The Lamp, has described AI as an evil that people should reject. Marc Barnes of New Polity argues that AI chatbots are themselves evil because conversational interaction with something that is not a person involves the user in an irrational act. He has even called for Magisterium AI, the Catholic platform built around Church documents, to be deleted. Other Catholic critics regard AI-generated sacred art as a dangerous intrusion into Catholic worship and culture.

And Eliza Monts, one of the movement’s most visible advocates, compares these early days of Catholic resistance to AI with what she has read about the beginnings of the American pro-life movement. She points to mounting public opposition to data centers and generative AI as evidence that resistance is beginning to coalesce.

Based on what I have already written regarding AI and Catholicism, I think such a position deserves attention, and there is absolutely nothing wrong with a Catholic deciding that AI is not good for him or her.

A person may conclude that ChatGPT weakens his concentration, that generative images offend his artistic sensibilities, that smartphones damage his prayer life, or even that contemporary technology as a whole makes it more difficult for him to live virtuously. He or she is free to walk away.

Indeed, Catholicism possesses a long and beautiful tradition of radical renunciation. The hermit leaves goods behind that are perfectly legitimate for others because he believes his own vocation requires something different. Monks voluntarily renounce possessions, marriage and aspects of ordinary social life—not because those things are evil, but because renunciation can itself become a path toward God.

In other words, families can decide to live without smartphones. I personally know and love many of them. My two favorite godsons belong to such families. Also, writers can refuse AI and artists can use only physical media. Such decisions may cultivate attention, craftsmanship, silence, presence and human communion.

But as much as Catholicism has plenty of room for technological ascetics, it does not have a theological basis for converting a personal ascetical choice into a universal moral prohibition.

That is an entirely different claim.

Here is where I part company decisively with the emerging anti-AI movement. A Catholic may reasonably say: I believe using conversational AI would be harmful to my intellectual or spiritual life. He or she cannot therefore conclude: Your use of AI is sinful.

Nor can he or she establish that Catholics attempting to integrate artificial intelligence into education, medicine, scholarship, evangelization or Catholic intellectual life are somehow betraying the faith simply because they have chosen engagement rather than abstinence.

There is currently no such teaching of the Church. More importantly, the direction of Pope Leo XIV’s teaching points elsewhere.

Magnifica Humanitas contains extensive warnings about artificial intelligence, but the encyclical explicitly refuses to frame the choice as one between enthusiasm and fear. Instead, Leo speaks of human creative intelligence as a gift capable of alleviating suffering and opening new possibilities and says that genuinely humanizing technological development can be embraced responsibly.

And Pope Leo made the same distinction even more clearly this past Friday, August 21, addressing Catholic legislators gathered to discuss “Human Dignity and Artificial Intelligence.”

Their task, he told them, is “not to hinder innovation but to guide it wisely.” Sound law should encourage technological creativity while protecting human dignity, rights, privacy and democratic institutions.

That is essentially a two-pronged Catholic approach: Beware and engage.

There is another irony here that is difficult to ignore. Some of the strongest Catholic critics of artificial intelligence quoted in the Register’s article are themselves extremely sophisticated users of digital media.

They publish online, build audiences on social networks, master the ever-changing algorithms of social media, produce podcasts and videos and distribute essays instantly around the world using technologies that would have been almost unimaginable a generation ago.

There is nothing wrong with that. I do exactly the same. But it raises an obvious question. Why did these technologies become morally acceptable while AI suddenly crosses some metaphysical Rubicon?

Christians expressed profound anxieties about the internet when it became widespread in the 1990s. And they were not entirely wrong. Pornography exploded. Human relationships changed. Attention became fragmented. Social media later produced new forms of addiction, manipulation and political tribalism.

Yet imagine if Catholics had concluded in 1995 that the appropriate response was simply to abandon the internet. No Catholic websites, no online catechesis, no myriad of ministries (from Catholic Answers to “Bible in a Year” to Catholic journalism distributed globally at negligible cost.)

The internet’s dangers did not disappear because Catholics participated in it. But Catholic participation made it possible for the Gospel to inhabit a cultural space that was going to exist whether Catholics approved of it or not.

The Church has repeatedly confronted technologies that radically changed human communication.

The printing press made possible the rapid dissemination of errors and heresies on an unprecedented scale. It also became one of the greatest instruments for spreading Scripture, theology, catechesis and Catholic culture.

Radio and television brought vulgarity, propaganda and commercial manipulation into the home. They also made possible extraordinary forms of evangelization. Just think of the upcoming beatification of Fultin Sheen.

The internet gave humanity pornography, misinformation and addictive social networks. But it also gave us instantaneous access to Scripture, the Fathers of the Church, papal documents, theological libraries and Catholic communities spanning continents.

Catholics who entered these new environments did not thereby endorse every consequence of the technology. They evangelized culture.

I know AI has its very unique challenges and complexities. That’s always the case with new technologies. In the end, I believe that history will settle part of this argument, and I do not mean that in the Hegelian sense.

Hegel’s famous formulation Die Weltgeschichte ist das Weltgericht (“world history is the world’s judgment”) is not my theology. History does not determine moral truth. Something does not become good simply because it triumphs technologically or culturally.

What I mean is that history teaches. And it teaches repeatedly that technologies possessing enormous practical usefulness are extraordinarily difficult to abolish. Attempts at complete cultural rejection usually fail. The more important question becomes who shapes the technology, which values become embedded within it and toward what ends it is ultimately directed.

That is where I believe Catholic energy should be concentrated. We need Catholics studying, criticizing and producing AI.

What we do not need is an invented doctrine according to which refusing AI represents fidelity while engaging it represents compromise.

Especially when the actual Magisterium is saying something rather different. On August 21, Pope Leo offered Catholic legislators what may be the clearest concise formulation yet of his approach:

“Dear friends, the world does not need an artificial intelligence that diminishes humanity; rather, it needs human intelligence enlightened by wisdom, political authority guided by conscience and technological innovation directed by charity. Only then will artificial intelligence become not a rival to humanity but a servant of the common good.”

That is the Catholic position I can wholeheartedly join: going where the culture is and trying to bring Christ with them.

The title of Hollis Robbins recent article “Universities would prefer not to” borrows from Herman Melville’s short story “Bartleby, the Scrivener: A Story of Wall Street” (1853,) which tells the story of a quiet clerk in a New York law office who slowly stops working and starts responding to every request with the polite phrase “I would prefer not to”.

In the case of universities, prefer not to change, that is.

Robbins’s argument is deliberately harsh. Four years after ChatGPT shocked academia, she says, much of American higher education has become less an accelerator of society’s adaptation to AI than a brake on it. Frontier AI research has largely moved into industry, while universities, the institutions supposedly devoted to knowledge, discovery, and preparing the next generation, remain preoccupied with preserving curricula, assessment systems, and institutional habits formed before generative AI existed.

Don’t get me wrong. Universities are right to worry about intellectual honesty. A student who asks an AI system to write an essay or a scholar who fabricates research commits a genuine offense against truth. Catholic education in particular should take those concerns seriously.

But protecting intellectual integrity is one thing. Protecting the old academic machinery is another.

And I agree with Robbins suspicion that some of higher education’s resistance to AI has considerably more to do with the second.

As we all know, universities have spent recent years suffering a serious loss of public confidence. Gallup measured confidence in higher education at 57 percent in 2015. In 2026 it stands at only 38 percent. Pew found last year that seven in ten Americans believe higher education is moving in the wrong direction.

As Robbins explains, higher education can appear remarkably insular: confident in its own procedures, protective of its credentialing power, and frequently more interested in defending the structures it already controls than asking whether those structures still serve students and society well.

Robbins offers a wonderfully instructive historical analogy. In nineteenth-century Britain, the Post Office controlled telegraph communications. When Alexander Graham Bell’s telephone appeared, one senior engineer dismissed its usefulness partly because Britain already possessed a “superabundance” of messenger boys. The result was not that the telephone disappeared. Britain simply adopted it more slowly, for everyone’s loss except the chosen bureaucracy.

There is a lesson here for universities: AI is being used everywhere around them and within them. In other words, a university can forbid ChatGPT in an essay assignment, but it cannot forbid the world into which that student will graduate.

Much of academia still seems focused on the question: How do we stop students from using AI to do what we have always asked them to do? They are avoiding a more interesting one: What should we now ask students to do because AI exists?

If AI can provide information cheaply and instantaneously, then professors should spend more time bringing students toward what Robbins calls the frontier; the areas where knowledge is incomplete, contested, local, experiential, or genuinely new.

That would not only not lower academic standards, but might actually raise them, and would return universities to what they claim to be: institutions at the frontier of knowledge rather than guardians of yesterday’s methods for transmitting it.

There is also something distinctly Catholic at stake here. The university, historically, is one of the great achievements of Christian civilization. At its best it is not a credential factory. It is a community ordered toward truth, where intellectual formation occurs through reason, discipline, argument, mentorship, and the integration of different branches of knowledge.

That conception of education should make Catholics less frightened by AI, not more. That’s because if a machine can save a lot of time and energy in the process of knowledge-wathering, then perhaps the human parts of education can become more important.

AI could potentially open more time and energy for mentorship, the patient examination of difficult questions, the formation in virtues and the integration of knowledge with moral responsibility. Those are things that were at the core of higher education, but have slowly disappeared.

Robbins reports that a large share of faculty and students in the United States and Canada remain unusually receptive to institution-wide AI bans, while only 17 percent of students in the region believe their instructors are well equipped to guide them in using AI. She describes most institutions as existing in a “state of AI paralysis.”

That is not sustainable. Reasonable guardrails should protect learning, not institutional inertia. The great irony would be for institutions dedicated to advancing knowledge to become obstacles to one of the most important knowledge technologies of our lifetime.

A perspective based on basic realism should instead bet on teaching students how to use AI without becoming dependent upon it, and recognize the difference between information and wisdom.

But for that to happen they have to stop preferring not to.

In my previous post, I argued that the future of artificial intelligence will require something more sophisticated than the slogans “accelerate” or “slow down.” Some areas may indeed require stronger guardrails and even deliberate pacing. But others may demand exactly the opposite.

Medicine is one of them.

A recent Wall Street Journal report offers a remarkable glimpse of what artificial intelligence is already doing for patients with rare and difficult-to-diagnose diseases. These are often people who spend years moving from doctor to doctor without an answer, sometimes being told that nothing serious is wrong while their condition continues to progress.

AI is beginning to shorten that diagnostic odyssey.

One mother, Rachel Hinken, had spent years wondering why her son Oliver was unusually small and missing important developmental milestones. Physicians repeatedly reassured her that he would catch up. Eventually she uploaded his photograph to Face2Gene, an AI-assisted tool that uses facial characteristics to identify possible genetic disorders.

The system suggested a rare condition called Trichorhinophalangeal syndrome. Genetic testing later confirmed not only that her son had the disorder, but that she had it as well.

After years of uncertainty, an algorithm helped point doctors toward the answer.

Other examples are equally striking. Physicians treating POTS, a difficult neurological condition, report that patients who previously might have gone decades without a diagnosis are increasingly arriving at specialists after AI chatbots helped them recognize patterns in their symptoms.

From a Catholic perspective, this is exactly the kind of artificial intelligence we should want to develop faster. Medicine itself is one of the clearest examples of human reason placed at the service of life. Hospitals, nursing, medical missions and care for the sick have occupied a privileged place in Christian civilization because healing the suffering person is not merely an economic activity. It is an act of mercy.

If AI can help physicians see what they previously could not see, we should ask how to make it safer, better and more widely available.

Because a technology capable of identifying a life-threatening disease months or years earlier should not be trapped behind unnecessary barriers simply because AI, in other fields, presents genuine dangers. Prudence does not mean applying the same brake everywhere.

Sometimes prudence requires acceleration.

The Journal article also contains the necessary warning. These systems still make mistakes. They are constrained by the quality of the data on which they are trained, and rare diseases by definition have fewer cases, fewer images and fewer digitized records available.

One clinical geneticist quoted in the article makes the proper distinction: AI can search across far more information, far faster than any individual physician, but human beings remain necessary for synthesis and final judgment.

And there is something almost beautiful about that. The more AI assumes some of the enormous burden of searching medical literature, recognizing complex patterns and processing diagnostic information, the more medicine could potentially recover what bureaucracy and modern healthcare systems have often squeezed out of it: time for the patient.

There are certainly areas where frontier AI capabilities may require carefully designed brakes. Autonomous cyber systems, biological risks, recursive AI development and other potentially catastrophic applications deserve serious oversight.

So, regulation has two moral dangers: allowing dangerous technology to advance recklessly, and preventing beneficial technology from reaching people who need it.

Catholic prudence must be capable of seeing both.

Let AI analyze scans and ECGs for patterns no physician has encountered before; but let the doctor remain the doctor: the person who verifies, judges, explains, accompanies and accepts responsibility.

That is a division of labor worth encouraging.

A few weeks ago, I wrote here about the remarkable public appeal from more than a thousand employees of America’s leading AI companies asking governments to develop the capacity to “pace” frontier artificial intelligence.

I welcomed the concern but remained skeptical. The basic problem seemed obvious: asking OpenAI, Anthropic, Google DeepMind and their international competitors to slow down together sounded morally admirable but practically unrealistic. These companies are engaged in one of the fiercest technological races in modern history. Add China to the equation, and the danger becomes even clearer: badly designed regulation could end up restraining the most responsible Western companies while rewarding whichever competitor decides not to play by the rules.

Even worse, “slow down AI” can become a dangerous political slogan. Once governments acquire broad powers to decide which technologies may proceed, there is no guarantee those powers will remain limited to genuine catastrophic risks.

Now Tim Fist and Saif Khan of the Institute for Progress, joined by several other analysts, have offered what I think is a much more realistic way of approaching the problem.

Their essential insight could be described as guardrails for the guardrails.

They do not reject the possibility that AI development may eventually need to be deliberately paced. Nor do they dismiss the increasingly serious concerns coming from people inside frontier laboratories. Instead, they ask a much harder question: If we ever need to slow something down, how do we make sure the mechanisms designed to protect us do not themselves become destructive?

That is exactly the question that was missing from much of the original discussion.

Fist and Khan propose that any intervention should be narrowly focused on activities carrying the possibility of serious and irreversible harm. It should minimize interference with the spread of already useful AI. Ideally, it should actually accelerate beneficial uses. It should remain worthwhile even if some of today’s more frightening forecasts turn out to be exaggerated. It should not punish the companies or countries acting most responsibly. And, crucially, it should not create a regulatory apparatus easily captured by a handful of corporations or misused by whichever political party is running the government.

We have spent time and the proverbial “ink” talking about guardrails for AI. We should now be talking about guardrails for the people who build the guardrails.

From a Catholic perspective, this is simply realism about power. Political authority is necessary. Markets cannot regulate every danger on their own, especially when powerful incentives encourage companies to move faster than their competitors. But the government is not morally immaculate either. Institutions can be captured and “emergency” powers can become permanent powers.

Catholic social thought has never required us to choose between blind faith in the market and blind faith in the state. It asks instead whether power is being exercised prudently, proportionately and in service of the common good. Or in the words of a famous German theologian: “us much society as possible, as much government as necessary”

That is why the idea of “low-regret” AI policy is so attractive.

Rather than announcing a sweeping pause today, Fist and Khan propose preparing institutions for the possibility that certain thresholds may eventually be crossed. For that, they propose:

• Build better transparency.

• Improve the government’s technical competence.

• Develop verification systems so that agreements can actually be enforced.

• Strengthen cybersecurity and biosecurity.

• Establish specific risk thresholds in advance.

If those thresholds are crossed, redirect some computing power and scientific talent away from the most dangerous forms of automated AI research and toward safer research and useful applications.

That is a much more intelligent version of pacing. In fact, slowing down one narrow area of AI development might allow the benefits of AI to spread faster elsewhere.

Compute and talent diverted from risky recursive self-improvement could instead be used for medicine, scientific discovery, cyberdefense, education and countless commercial applications that already exist.

This is not a choice between brakes and acceleration. A well-designed vehicle needs both.

But prudence does not mean proceeding slowly simply because slowness feels safer. Nor does it mean charging ahead because progress is exciting. Prudence asks what the circumstances actually require and then chooses the appropriate means to achieve a genuine good.

Sometimes prudence says accelerate, sometimes it says brake… and sometimes it says build better brakes precisely so that you can drive faster without killing anyone.

The AI debate desperately needs that kind of thinking.

And perhaps that is the emerging lesson of the AI debate: we need guardrails for artificial intelligence, but we also need guardrails for the guardrails. Because both technological power and regulatory power are human powers; and neither should ever be trusted without limits.

Mark Zuckerberg has entered the increasingly crowded field of AI manifestos with an ambitious declaration titled The Future Is for Everyone. It is long, optimistic, sweeping… but at times so sugary that one begins to suspect that Meta’s public-relations department was serving dessert.

Still, beneath all the rhetoric there is an argument worth taking seriously.

Zuckerberg’s case can essentially be reduced to basic economics and political realism: concentrated power produces monopoly-like advantages, while distributed capability encourages competition and spreads those advantages throughout the economy.

Applied to artificial intelligence, his argument is straightforward. If only a handful of governments or corporations possess extremely powerful AI, those institutions will acquire enormous advantages over everyone else. If, instead, increasingly capable AI becomes cheap and widely accessible, millions of individuals, entrepreneurs, researchers, and small companies will be able to use the same technology to compete, invent, and create.

That part of the manifesto makes sense.

Indeed, from a Catholic perspective, there is something attractive about it. Catholic social teaching has always been suspicious of excessive concentrations of power, whether political or economic. Subsidiarity favors empowering persons, families, communities, and smaller institutions rather than automatically transferring power upward to gigantic centralized structures.

But then comes the frosting.

Much of the manifesto wanders through optimistic descriptions of personal empowerment, creativity, relationships, health, education, community benefits, individual aspiration, and a future in which almost everyone becomes more capable of realizing his dreams. Some of this may prove true. But page after page of technological sunshine does little to answer the anxieties ordinary people actually have about AI. Most importantly, it ignores the proven dark side of humanity, something Catholics call “original sin.”

Zuckerberg argues that there will ultimately be “an abundance of jobs,” that AI will create occupations we cannot yet imagine, and that companies may become smaller while the total number of companies increases.

I hope he is right.

But that is less an answer to the employment problem than a bet on human economic adaptation.

The question haunting millions of workers is not whether economists can imagine new occupations appearing eventually. It is what happens between the disappearance of today’s job and the arrival of tomorrow’s hypothetical one. In other words, for a significant number of today’s workers, it is all about the transition.

A Catholic approach cannot judge technological progress merely by whether aggregate productivity eventually rises. The worker is not an adjustment cost. The time between technological displacement and economic reinvention is lived by actual human beings with mortgages, marriages, children, obligations, fears, and dignity.

And then there is Zuckerberg himself.

This is perhaps the manifesto’s greatest practical weakness: the messenger.

Zuckerberg wants Americans to believe that Meta has suddenly emerged as the champion of distributing power away from large institutions and toward ordinary individuals. One can appreciate the intellectual argument while recognizing how extraordinary that sales pitch sounds coming from the founder of one of the most powerful technology corporations in human history.

To be brutally clear: A 2025 Pew survey found that roughly two-thirds of Americans viewed Zuckerberg unfavorably, while Meta’s history includes the FTC’s record $5 billion privacy settlement; and as recently as August 2026, 61% of Americans told Reuters/Ipsos they favored stronger oversight of social-media companies. .

Therefore Meta surely understands why some people might raise an eyebrow… or both.

This becomes especially problematic when the manifesto adopts an almost philanthropic tone. Zuckerberg may sincerely believe that personal superintelligence will empower billions of people. I have no reason to claim otherwise. But sincerity is not the same thing as credibility.

This is where Catholic realism becomes particularly useful. We should judge Zuckerberg’s proposal neither according to our affection nor our dislike for Zuckerberg. The principle itself deserves consideration: distributing AI capabilities broadly may indeed be preferable to allowing a tiny oligopoly of corporations or governments to control the most powerful technology of our age.

But Catholic realism also reminds us that economic power does not cease being power merely because those possessing it speak the language of empowerment.

“The Future Is for Everyone” cannot merely mean that everyone gets access to the machine.

It must also mean that the economic order surrounding the machine remains genuinely ordered toward everyone.

That is the part of the manifesto that still needs to be written.

One of the most important insights in the launch of Pax Machina is also one of the easiest to overlook: if artificial intelligence is truly revolutionary, then we cannot assume that the institutions governing society can remain essentially unchanged.

That sounds obvious, but much of the AI debate still focuses almost exclusively on the technology itself. Should we slow it down? Accelerate it? Regulate it? Align it? Make it safer?

Pax Machina asks a different question: what kinds of institutions will be necessary in a world shaped by powerful AI?

The editors point out that previous technological revolutions transformed institutions as much as they transformed machines. Industrialization, for example, helped produce the modern corporation, labor unions, regulatory agencies, and mass public education. The internet revolution generated new forms of markets, reputation systems, governance, communication, and coordination.

Powerful AI will almost certainly do the same.

Existing checks and balances may prove inadequate when AI systems can operate at speeds and scales for which those institutions were never designed. Governments may gain extraordinary administrative powers. Courts, elections, media, diplomacy, and markets may all function differently when AI agents participate in them directly or when institutions themselves begin using AI to make decisions.

Pax Machina’s central intuition is therefore exactly right: institutional change is coming whether we design it deliberately or simply allow it to happen.

From a Catholic perspective, however, there is one indispensable addition. Institutional innovation cannot be reduced to mechanism design.

In other words, it is not enough to ask whether a new institution is efficient, responsive, scalable, or technically elegant. We also have to ask what vision of the human person it embodies.

Does it protect genuine human agency or quietly replace it? Does it strengthen subsidiarity or concentrate power? Does it protect families, workers, and the vulnerable? Does it preserve moral responsibility when decisions are mediated by machines? Does it recognize that efficiency is not the highest social good? And does it leave room for conscience, freedom, solidarity, and the irreducible dignity of the person?

Pax Machina itself recognizes the danger. The same AI-enabled institutions that could make government dramatically more responsive could also make surveillance and political manipulation far more powerful.

That is precisely why Catholics should welcome this conversation and participate in it.

The Church has a long institutional memory. Universities, hospitals, charitable networks, religious orders, canon law, guild traditions, and countless forms of association grew from the conviction that institutions should not merely coordinate people efficiently. They should help order human freedom toward genuine goods.

Pax Machina is right that we should begin designing those institutions now. But the drafting table needs more than engineers, economists, lawyers, and political theorists. It also needs moral anthropology.

Otherwise, we may succeed brilliantly at designing institutions for powerful AI while forgetting to design them for human beings.

Geoff Shullenberger (left); Eliezer Yudkowsky (right) (images selected by Grok)

Geoff Shullenberger’s recent Washington Post essay may be one of the clearest summaries yet of the current metaphysical state of the AI debate.

As we know, the public discussion is currently awash with pragmatic takes on AI. Will it destroy jobs? Will it cure disease? Will it make companies more productive? Should governments regulate it? Those questions are important, but they no longer reach the deepest level of what is happening in Silicon Valley.

Shullenberger’s argument is that AI has pushed the technology world beyond questions of means and into questions of ultimate ends. Technology used to ask principally, “How can we do this?” The AI movement increasingly asks, “What should humanity become?” And once that happens, the debate inevitably becomes theological.

He traces this development through figures such as Eliezer Yudkowsky, whose early writings did not describe superintelligence merely as a useful invention but as an “Interim Meaning of Life” and even a “temporary definition of Good.” In other words, the early Yudkowsky, a strong believer in radical neo-rationalism, argued that AI ceased to be a tool serving human purposes and became itself the purpose toward which humanity is ordered.

From there, the religious parallels multiply. There are prophets announcing the coming transformation. There are rival eschatologies: salvation through superintelligence on one side -say, Peter Thiel’s side- extinction through superintelligence on the other -Yudkowsky’s current pessimism-.

There are heresies, schisms, competing sects, Antichrist accusations, visions of immortality, and even arguments that the attempt to create nonhuman intelligence resembles occult invocation.

Shullenberger’s central insight is that both the AI accelerationists and many of the AI safety advocates share an underlying premise: humanity is destined to create an intelligence greater than itself. They disagree violently about whether that event will save us or destroy us, but they often agree that it represents the decisive horizon of human history.

This is why Pope Leo XIV’s intervention is so significant. As Shullenberger notes, the Pope does not merely condemn the technological impulse. He recognizes even in the crazy dreams of “transhumanism” a distorted but recognizable human yearning for “a fuller life.”

Christianity, indeed, does not mock the longing for transcendence. It says that the longing is real; but that its fulfillment cannot be manufactured in silicon.

So, the Christian disagreement with technological messianism is much deeper than an argument over safety standards. Catholicism insists that intelligence is not the highest good, that human limitations are not engineering defects, and that salvation cannot be achieved by constructing a superior mind. Man is not an unfinished machine waiting to be upgraded. He is a creature made in the image and likeness of God, wounded by sin, called to communion, and destined for a fulfillment no technological system can provide, because it was brought by Christ.

Shullenberger ends with an important warning. The current rapprochement between religion and Silicon Valley does not necessarily mean religion will shape technology. The influence may flow in the opposite direction.

If religious institutions engage with AI without recognizing those assumptions, they may discover that they are being evangelized by Silicon Valley rather than evangelizing it.

In other words, despite the evident interest of powerful AI leaders for religion, Catholics should not mistake spiritual curiosity for conversion.

The emerging AI culture already possesses something resembling a theology: its own doctrine of human limitation, its own eschatology, its own promised transcendence, and its own competing visions of salvation and apocalypse.

The task of the Church is not simply to add Christian ethics to that technological worldview. It is to challenge the worldview at its roots.

There is something wonderfully entertaining about visiting the online Archive of Incorrect AI Predictions. It is a museum devoted to a persistent human talent: confidently predicting the future… and then watching the future refuse to cooperate.

The exhibits are especially amusing because the mistaken forecasts did not come chiefly from fools or amateurs. They came from some of the most brilliant, informed, and influential people in computing.

In 1958, Herbert Simon and Allen Newell predicted that within ten years a computer would become the world chess champion. They were only off by about three decades: Deep Blue finally defeated Garry Kasparov in 1997.

In 1970, Marvin Minsky predicted that within three to eight years we would have a machine with the general intelligence of an average human being. The deadline arrived. The average human being remained stubbornly unreplaced.

In 2016, Geoffrey Hinton famously suggested that people should stop training radiologists because deep learning would outperform them within five years. Instead, radiologists remain very much employed, often using AI as an aid rather than being replaced by it.

More recently, Sam Altman acknowledged that he had expected AI to eliminate entry-level white-collar jobs more rapidly than it actually has. He had been broadly right about the technology’s advancing capabilities, he admitted, but wrong about the speed of its social and economic consequences.

The lesson is not that artificial intelligence is incapable of producing enormous disruption; it is rather simpler and more humbling: even the people who understand a technology best are often remarkably bad at predicting when its consequences will arrive and what human beings will do in response.

This should bring some relief to Catholics trying to navigate the current atmosphere of AI enthusiasm and panic. The debate has developed its own secular eschatology. On one side, the optimists promise abundance, cures for disease, freedom from drudgery, and perhaps even the defeat of aging. On the other, the pessimists warn of mass unemployment, superintelligence, civilizational collapse, and machines slipping beyond human control.

Both sides may perceive something real. But both also face the same temptation: turning possibilities into certainties and projections into prophecies.

Catholic realism should make us suspicious of both technological salvation and technological apocalypse. Christianity does not teach that machines will redeem humanity. Nor does it teach that every new instrument of human ingenuity must become an engine of destruction. It teaches that human beings are creative and fallen, capable of directing powerful tools toward astonishing good and appalling evil—often at the same time.

Prudence lies precisely between complacency and panic. It studies what is happening, attends to concrete risks, makes reasonable preparations, and remains willing to revise its judgments as reality changes.

Most importantly, society is not a laboratory variable. Human beings respond, appropriate and transform technologies in ways their inventors rarely anticipate.

Thus, perhaps the most Catholic response to the AI prediction industry is a mixture of vigilance, humor, and humility. Because the safest prediction may be that nearly everyone is underestimating something and almost everyone has the timing wrong.

Recently, a Catholic publication announced that it had received an excellent submission -good enough, by its own admission, to publish- but rejected it after the author honestly disclosed that he had used artificial intelligence during part of the research process.

That decision strikes me as not merely excessive, but counterproductive. It punishes transparency while quietly encouraging less scrupulous writers simply not to disclose what tools they used. It also reflects a broader tendency among some Catholic editors and writers to treat any involvement of AI as a contamination of authorship; as though consulting a chatbot during research were morally equivalent to presenting an entirely machine-generated essay as one’s own.

These are plainly not the same thing.

Tyler Cowen confronts this anti-AI absolutism, with characteristic humor, in his recent essay for The Free Press titled “You’ll Learn to Love AI Writing”. He readily acknowledges that much current AI prose is irritating, predictable, and recognizable by its clichés. But he asks: “Do I wish to eschew AI writing for the rest of my life? Absolutely not.” What he wants is better AI writing: writing capable of escaping the obvious formulas that currently betray its origin. Indeed, he provocatively says: “I want AI writing that can fool me.”

That line will scandalize some readers. But Cowen’s deeper point is reasonable: the categorical rejection of AI-assisted writing is unlikely to survive once the technology becomes sufficiently good, convenient, and integrated into ordinary creative work.

And it is becoming integrated. Rapidly.

My own position is that especially serious, original writing, should remain primarily a human endeavor. The writer must still supply the idea, the judgment, the intention, the intellectual architecture, and the final responsibility for every sentence published under his or her name. AI should not become a mechanism for manufacturing opinions one has never seriously considered or arguments one could not personally defend.

But that does not mean refusing every form of assistance.

I use AI extensively to help draft ideas and, especially, to polish my English. English is not my mother tongue, although professional circumstances require me to write in it frequently. If AI can help me express my own thoughts more clearly, precisely, and elegantly for the reader, I see no betrayal of authorship in that. The thought remains mine, as well as the decision about what stays and what goes. Most importantly, I remain responsible for the result.

An editor does not become the author of a book by improving its prose; nor does a translator erase the original author by finding better words in another language. Spell-checking software does not invalidate an essay. Research databases, dictionaries, search engines, transcription programs, and grammar tools have already become normal parts of writing.

Granted, AI is more powerful and morally complicated than these tools, but power and complexity demand discernment rather than automatic condemnation.

Creativity Has Never Meant Creation From Nothing

Cowen illustrates his argument with Paul Simon’s “American Tune.” Its celebrated melody is closely associated with a chorale used repeatedly in Johann Sebastian Bach’s St. Matthew Passion. Yet Bach himself did not create the underlying melody. It originated around 1600 as a secular love song by Hans Leo Hassler, was later adapted for Christian use, and eventually received Bach’s magnificent harmonization before being repurposed centuries later by Simon.

Does that make Paul Simon or Bach, for that matter, plagiarists?

Obviously not. It shows that human creativity is often cumulative. Artists inherit, adapt, rearrange, translate, answer, and transform what came before them.

Genuine creativity does not always mean producing something without precedent. Strictly speaking, only God creates ex nihilo. Human beings create as creatures: receiving a world they did not make and working with materials, languages, traditions, melodies, images, and ideas inherited from others.

Catholicism should understand this better than almost anyone. The Catholic intellectual tradition is built upon reception and development: Scripture and commentary, faith and reason, ancient philosophy and Christian transformation, inherited forms continually applied to new historical circumstances.

Of course, that does not abolish plagiarism. Plagiarism occurs when a person deceptively claims another’s intellectual work as his own. AI introduces difficult questions because its sources may be opaque and because the line between assistance and substitution can become blurry. But those difficulties cannot be resolved by declaring that any AI involvement destroys authorship.

Thus, the moral question is not simply “Was AI used?”

The serious questions are: Who originated the central idea? Who exercised intellectual control? Who verified the evidence? Who made the decisive creative judgments? Was the machine’s contribution substantial enough that disclosure was reasonably owed? Were readers deceived? Were sources fabricated or misrepresented? And who accepts responsibility for the final work?

Those questions require prudence. A blanket prohibition requires almost none.

What a Catholic Standard Should Look Like

A sensible Catholic approach should defend several principles.

First, no writer should use AI to fabricate facts, quotations, experiences, or sources. Second, no one should present machine-generated work as the fruit of intellectual labor he did not perform. Third, the author must verify factual claims and remain accountable for every word bearing his name. Fourth, material AI involvement should be disclosed when its concealment would give readers or editors a false impression about how the work was produced.

But disclosure should be proportionate. A writer should not need to append a technological confession because AI corrected grammar, suggested a clearer sentence, generated possible headlines, transcribed an interview, or helped locate research leads. Requiring that would be both absurd and unenforceable.

The decisive issue is whether AI has assisted human authorship or replaced it.

That distinction is entirely compatible with Pope Leo XIV’s Magnifica Humanitas. The encyclical does not frame the challenge as a simplistic choice between accepting or rejecting technology. It insists that technology is never neutral because it bears the imprint of those who design, finance, regulate, and use it. It therefore calls for shared discernment and teaches that education must help people determine not only how to use AI, but when and for what purposes it should not be used.

That is a far more demanding approach than technological phobia. It asks Catholics to judge specific practices according to truth, justice, human dignity, responsibility, and the common good.

The Luddite Option Is Not an Option

The absolutist position is also historically unserious. Technologies that offer dramatic convenience do not disappear because cultural guardians disapprove of them. The printing press did not retreat before defenders of the manuscript: even if those defenders included a genius like Johannes Trithemius. The internet was not stopped by editors who preferred paper. Search engines did not vanish because research once required hours in a library.

Some old crafts were weakened. Certain disciplines were lost. New temptations appeared. But the technologies remained because they expanded human possibilities too powerfully to be ignored.

AI will be no different.

Catholic publications can establish intelligent standards. They can insist upon disclosure where it genuinely matters, punish deception, demand factual verification, preserve human accountability, and protect the distinctive value of human testimony and style. What they cannot realistically do is pretend that serious writers will never use AI, or that every writer who does has somehow betrayed the human vocation to create.

That attitude will not preserve human authorship. It will merely drive AI use underground while making Catholic institutions appear frightened, confused, and technologically unserious.

The right Catholic response is neither surrender nor prohibition. It is disciplined collaboration. AI may assist research, challenge an argument, identify weaknesses, improve organization, translate ideas, and polish language. But the human author must remain the originator, judge, moral agent, and responsible bearer of the work.

This boundary will not always be easy to draw. As AI improves, some cases will become more difficult, not less. We will need policies, professional norms, transparency, and continued moral reflection. But difficulty is no excuse for an infantile binary in which unaided writing is pure and any technological assistance is corrupt.

The history of culture has always involved tools, inheritance, adaptation, and collaboration. AI is a new and unusually powerful participant in that history. The task before Catholics is not to deny that reality, but to enter it with confidence, prudence, and moral clarity.

Human authorship does not require the writer to work without tools. It requires that the tools remain ordered to a genuinely human act of intelligence, freedom, creativity, and responsibility.

Something remarkable is happening inside the artificial-intelligence industry. More than a thousand employees of the world’s leading AI companies, including senior researchers, chief scientists and several co-founders, have signed a public statement asking the United States government to support international efforts to “deliberately pace” the development of cutting-edge AI.

They are not demanding that artificial intelligence be buried or frozen indefinitely; but asking the US government and society in general to develop the technical and governing mechanisms needed to buy time if automated AI research begins accelerating beyond our ability to understand, secure or control the resulting systems.

These are not outside critics frightened by a technology they do not understand. Many are helping build it. And that’s why, from a Catholic perspective, the appeal has real value. Prudence is not timidity or hostility toward progress. It is the virtue that applies right reason to action, judging not only whether something can be done, but when, how and under what conditions it should be done.

The signatories are also acknowledging a fundamental problem of collective action. No company wants to slow down alone. Any laboratory that pauses to improve its safeguards risks losing talent, investment, customers and strategic advantage to a competitor that continues racing ahead.

The same problem exists among nations. The United States fears China; China fears the United States; smaller countries fear being permanently subordinated to whichever power reaches the frontier first.

This is why asking for international coordination makes sense. If advanced AI could eventually accelerate its own development, a purely voluntary pause by one company would accomplish nothing.

Meaningful pacing would require common standards, monitoring, verification and consequences for defection… and this is also why the proposal is so unlikely to succeed.

OpenAI, Anthropic, Google DeepMind, Meta and their emerging rivals are engaged in a brutal commercial contest. Their valuations, reputations and expectations of future dominance depend upon reaching the next breakthrough before everyone else.

The statement therefore reveals a striking contradiction. Many of the people building frontier AI want society to possess the ability to slow the race, while the institutions employing them remain economically committed to winning it.

Catholic realism should understand this tension immediately. Human beings can recognize the common good while still pursuing narrower interests. We can know that cooperation is necessary and nevertheless fear being the only ones who cooperate. Sin does not always appear as open malice.

That does not make the statement futile. Its first great contribution is that it establishes a fact which can no longer be dismissed: concern about uncontrolled acceleration is not confined to religious thinkers, philosophers or technological pessimists. It is present among the builders themselves.

Another important contribution is that it distinguishes pacing from permanent prohibition. Buying time can be pro-innovation, while allowing institutions to prepare for economic disruption and society to expand the beneficial uses of AI without recklessly multiplying its dangers.

Of course, any credible framework will require verification, transparency and accountability—not merely admirable promises.

Because the alternative is worse. If every company and country concludes that restraint is impossible because nobody trusts anybody else, then the race itself becomes sovereign and human beings cease governing technology and begin obeying the imperatives created by competition.

Creating that space will require more than signatures. It will require companies and nations to accept that victory in the AI race is not worth achieving if, in the process, humanity loses the ability to govern what it has built.