Viktor Poletko (PhD, KU Leuven) is a political theorist and head of business unit at an international software consultancy.
This post is a part of our series on Magnifica Humanitas.
The idea that this Pope would kick off the contribution on social ethics was pretty obvious. Given the papal name he chose for himself, the theme of his first encyclical was easy to guess. But while Pope Leo XIII famously addressed the societal and economic effects of the Industrial Revolution, Leo XIV turned his attention to an allegedly overhyped topic of the AI revolution.
The Fourth Revolution
Magnifica Humanitas (MH) speaks of a “fourth industrial revolution”—Klaus Schwab’s term—placing AI alongside the earlier ruptures of steam engine, electrification, and digitalization (para. 151). Based on this reading, what is at stake is nothing less than a fundamental reordering of our social and economic life. The document holds that technological development is rapidly reshaping “languages, relationships, institutions and forms of power” (para. 90). Nonetheless, it is important to stress that the encyclical’s overarching view of information technology is favorable. It not only acknowledges that technology “can greatly serve integral human development and the care of our common home” (para. 93) but also, more specifically, that it has a desirable role in “relieving humans of arduous, repetitive or dangerous tasks” and rendering “intelligent support for human activity” (para. 152). Yet the document also recognizes that technology carries immense power. And inasmuch as it carries power, it also poses risks.
Those risks and challenges posed by technological changes are many—from the very specific concentration of socioeconomic power in the hands of commercial players owning technological assets, to the more general lagging of moral progress behind technological progress. In what follows, I focus on just how impactful, on a close reading of the encyclical, the emergence of AI has become, both for the human-rights landscape and for the business sector.
Power over Rights
A familiar distinction in legal and political theory separates two families of rights: liberty-rights and equality-rights. This post discusses AI–related risks pertaining to each family and the remedies proposed by Magnifica Humanitas to address those risks.
Regarding liberty, the first concern relates to the loss of inner freedom, as individuals fall victim to addictive, attention-tracking digital techniques (MH, para. 170). Digital technologies commodify users’ time and attention, thereby instrumentalizing them—objectifying users and turning them into a means exploited for commercial ends. The second concern is even more straightforward. Thanks to massive data collection, virtually every parameter of human life can be measured, and as managers well know, what can be measured can also be shaped—that is to say, quantification enables control. The exercise of this power is subtle but very palpable, working primarily through self-censorship and the pressure to conform to behavior deemed acceptable by state bodies or commercial giants.
Equality, too, is under siege—or at least at risk of siege—on several planes. As the encyclical emphasizes, over-dependence on technology widens the gap between those who have equal access to resources and opportunities and those who do not (para. 80).
Even more importantly, the very fundamental problem of AI—namely, the skewed data used for model training—creates an unavoidable bias and therefore undermines epistemic equality, as the documented cases of skewed or discriminatory model outputs demonstrate. Not to mention the deliberate poisoning of data—artificially seeding false content so that it ends up in training sets and skews what models later say.
And worse still, delegating decision-making power to an algorithm not only risks discrimination for the very same reason but also diminishes, if not removes, the moral sensitivities and political responsibility that ought to rest on the shoulders of the relevant human actors.
This is why AI is not to be regarded like other, neutral tools, such as a pen, a hammer, or money, which take on a moral character only once they are used for good or evil—and even then, only conditionally. By contrast, the innate biases of AI constitute, according to the encyclical, an embedded moral character that it bears from the very moment of construction (para. 104).
This is by far not an exhaustive list of the threats outlined in the document, but I believe it captures the spirit of what is at stake. It is precisely these human-rights concerns that invoke caution about progress. Now, what to do about them?
Toward Remedies
The encyclical’s answer is straightforward: we have to disarm AI—by regulating it at the policy level, by detaching it from the logic of armed competition, and by being more conscious of the values conveyed when technology is designed and developed. But since neither activity is risk-free, the encyclical calls for strengthened political involvement and democratic deliberation. Despite only a brief mention, this participatory aspect may be one of the most important theses of the whole document.
Specifically, with regard to liberty-related concerns, the Pope proposes measures that build up inner freedom—educating toward digital restraint, protecting minors, and countering business models that exploit human vulnerability (para. 170). Furthermore, beyond these measures aimed at inner freedom, a strong call is made for concrete policy-level regulation of intrusive technologies—however broadly the term intrusive is to be understood (para. 171).
Moving to equality, the call is made not only to analyze the purposes behind any given AI system but also to scrutinize exactly how it is designed, and which specific underlying conception of both individual and society is embedded in it (para.104). This is precisely to manage the embedded moral character of the technology. Further, just as the preservation of freedom requires individuals to account for how they use AI, so the preservation of equal treatment requires that specific individuals remain accountable for the “decisions” or, rather, the outcomes, produced by these systems (para.105).
A Policy Paper?
More often than not, what the encyclical proposes resonates with the standpoints taken by independent experts, governments, and even AI market leaders. Take, for instance, the recent essay by Dario Amodei, Anthropic’s CEO, on the necessity of strict regulation of the sector. Those proposals may be feasible to implement.
Yet that’s not always the case, especially when the remedies would necessarily touch on basic economic rights and principles.
In line with a classic argument reaching back at least to Michael Sandel’s work of more than a decade ago, technology as an asset, once commodified, deepens inequality and widens the social gap between those who have access to it and those who don’t. The argument holds—but how, exactly, to bring it to life in practice?
The same question applies to Pope Leo’s daring proposal to treat data as a common-pool resource (MH, para. 108). The moral desideratum makes sense, but how to make it work is far less clear. Excellent theoretical work has been done on the issue by a number of scholars, notably Elinor Ostrom, who showed how certain common-pool resources can be governed without falling into either heavy regulation or predatory privatization. Yet how to implement such a framework remains unclear—not least because those who profit from data as a private asset have little reason to give it up.
Proposing the mechanism, however, is not the purpose of the document. Encyclicals are moral compasses rather than policy papers. They are meant to signal problems and propose a vision, whereas carving out the actual path belongs to politicians, state bodies, and policymakers. As has always been the case, the more they fall into specific mechanisms of implementation, the more criticism they attract from the left or the right, or both.
Embracing Indeterminacy
One more issue worthy of attention relates to the assumptions made about the state of the art. While the future is by far not as apocalyptic as recent forecasts, such as “AI 2027,” would indicate, the encyclical still assumes we are in the midst of a huge disruption. This assumption is understandable. Our engagement with computational technology has grown to the point where it is perceived as essentially different from anything we have seen before. The media hype, amplified by financial-market enthusiasm, is mind-blowing. And yet, despite astonishing investment, optimism seems to have passed its peak.
In the day-to-day life of the private sector, we do see real transformations, but it would be fair to say that the challenges firms face now are very different from those anticipated just last year—from lowering AI infrastructure costs, to repairing the damage done by hasty decisions, to employing AI to perform narrow tasks in a commercially productive way. The chance that the AI revolution turns out to be just another gradual—but not fundamental—transition, without massive systemic consequences, is real and may not be small.
This uncertainty makes Magnifica Humanitas special. The encyclical starts by outlining the history of Catholic social teaching. It’s a show of and a call to continuity, which undoubtedly is present, as far as principles are concerned. But there is a difference in the position from which the teaching is offered. And the Pope is aware of that. He explicitly acknowledges that whatever one says about AI gets outdated in no time (para. 98). And this is where the radical difference lies.
Leo XIII foresaw in 1891 what unregulated capitalism would bring. John XXIII knew in 1963 what a global nuclear conflict would do to human civilization. John Paul II knew in 1991 that communism as a project had collapsed. Francis, too, sensed in 2015 what was happening with climate change—and, fairly clearly, where it was going to take us. But what is going to happen with AI? We have no idea. The moral guidance is there. But to what exactly does it apply, and what specific configuration of challenges to human rights will it pose one, five, or ten years from now? That remains to be seen.