AI, Pope Leo, and the Finns

1600 words, 7 minute read

AI sure makes strange bedfellows. Less than 1% of Finns are practicing Roman Catholics…much less than 1% in fact. That makes Finland the least Catholic country in all of Europe. And yet academics at the University of Eastern Finland collaborated in an article (9/8/2026) on AI Governance that is intentionally based on the teachings of Pope Leo XIV.

Our own commentary on the Pope’s first encyclical (Magnifica Humanitas) was less charitable. This was a ‘heaven sent’ opportunity for Leo to make a major contribution on the most important social issue of our time (AI); and in the process, it was a chance to reassert, and reestablish, the relevance of Roman Catholicism for the modern world.

In language a long suffering Chicago White Sox fan would appreciate, the Pope was ‘caught looking’, called out on strikes without ever taking the bat off his shoulders. Quelle domage! But wait, could Finland save Leo’s Papacy?

AI Governance ignores Leo’s Luddite commentary and concentrates on eight benign principles distilled from his encyclical: (1) human dignity, (2) the common good, (3) the universal destination of goods, (4) social justice, (5) solidarity, (6) subsidiarity, (7) shared responsibility, and (8) transparency/accountability. As we shall see, our authors place special emphasis on ‘universal destination’ and ‘subsidiarity’. All in, this is a welcome gloss on Leo’s document.

The authors review three current models for AI governance in light of those eight principles:

The State Model - The state is primary agent of governance. AI is a technology requiring regulation, centralized oversight, and enforceable mechanisms to protect civil rights, public safety, and the collective interest.

The Market Model - AI governance should be shaped primarily by its creators and stakeholders. This approach places emphasis on industry-wide standards, adopted codes of ethics, and the responsible behavior of technology producers.

The Hybrid Model – Governance requires the cooperation of multiple public and private actors. Responsibility is shared among public institutions, private companies, industry experts, civic organizations, and international bodies.

Leo’s reflections best align with the Hybrid Model and they add normative corrections to both state-led and market-led models. Human dignity and shared responsibility are prioritized over centralized control, public or private, all in pursuit of “a more just and inclusive social order in which technology enables human flourishing.” This is precisely the dimension of AI that Leo ‘underappreciated’.  

AI Governance points out weaknesses in all three models. For example, “centralization can lead to technocratic forms of control and reduced pluralism in which algorithmic systems restrict democratic deliberation and civic agency.” Amen to that!

But the authors follow Leo down the rabbit hole of Euro-American ‘democracy’. We are assured that “the state model in democratic societies ensures the protection of fundamental rights, public safety, and enforceable regulatory standards.” Apparently, the authors have never visited Planet Earth.

As the Pope’s namesake, Leo XIII (c. 1880) was quick to point out, the superficial form of a government does not always reflect its actual function. No paper constitution guarantees a just society. Did the US tolerate slavery; did the UK imprison homosexuals?  

The Market Model does not protect against the concentration of power in the hands of technocrats. Private sector governance often combines ethical declarations with ‘limited transparency, asymmetric power, and inadequate safeguards’. Public interest may be subordinate to the twin imperatives of competition and growth.

The Hybrid Model is championed as ‘best suited to the multidimensional nature of AI governance’. Ideally, it involves public and private interests, industry experts and civic actors, in a ‘polycentric power structure’. “However, institutional pluralism alone does not guarantee genuine accountability.” A hybrid structure may serve as a fig leaf for the “concentration of real control in the hands of dominant institutions or interest groups.”

AI Governance reimagines Leo’s encyclical as a reflection on the course of civilization itself. This brings us back to Leo’s eight principles (above). “Together, they form a vision of AI governance that stands in opposition to the logic of the over-concentration of power” and the reduction of human beings to the status of mere data points, i.e. means rather than ends.  

“In the context of AI governance, this means that the primary criterion for evaluating technology cannot be efficiency alone, but rather whether AI systems protect the human person from being instrumentalized.” Here is where Aletheia Today parts from Leo and the Leonides. The ‘human person’ has already been ‘instrumentalized’, if not by agrarian serfdom then certainly by industrial wage slavery!

Karl Marx wrote that the purpose of philosophy is not to interpret the world but to change it. Likewise, the potential of AI is not to ‘protect the human person’ but to free that human person from such instrumentality.

“From this perspective, AI governance cannot be limited to risk control but must also address equitable access to technological resources, data, infrastructure, and the benefits arising from the development of AI.”

Yes…but! AI (on top of PCs, smart phones, the internet, the web, and social media) is the engine of ‘equitable access to technological resources…and the benefits arising from the development of AI’. Just try to stop it! Job #1 for potential regulators: Don’t get in the way!

Our authors echo Leo’s concern that “the development of global networks, platforms, and AI systems may lead to new forms of exclusion, restrictions on freedom, invasive surveillance, and discrimination reproduced by opaque algorithms.” In fact, AI’s networks, platforms and systems are engines of inclusion, removing restrictions on freedom and protecting against discrimination.  (‘Invasive Surveillance’ remains a legitimate concern.)

Consider: “Discrimination reproduced by opaque algorithms.” AI doesn’t invent its own prejudices; it inherits them from its human trainers. AI can’t hate anyone unless one of us, consciously or not, hated that person, race, gender, or class first. Heck, we discriminate against red headed stepchildren; AI can only improve on our dreadful record!

But this is a teachable moment! We excoriate AI when it displays any sort of discriminatory behavior…and rightly so…but we fail to acknowledge that the prejudice originated with us. Let’s remove the logs from our own eyes before we worry about the splinters in AI’s. To hold AI to the highest possible ethical standards is a good thing; but let’s first hold ourselves to those same standards.

“A social order requires equal access to technological opportunities, protection of groups particularly vulnerable to marginalization, and public control over the use of data and technology: who possesses the data, computational capital, and capacity to train AI models, and who becomes merely the object of their operation.”

Of course…but let’s unpack. Again, AI is an engine of distributed access, not a barrier. That access protects those ‘vulnerable to marginalization’ and ‘public control’ is thereby assured, so long as we equate ‘public’ with ‘people’ and not the state. And in the long run, we will all ‘train AI’ as we live our lives and generate data in the process.

Leo and our authors correctly emphasize the importance of Solidarity, the recognition that “the future of each individual is connected to the future of all. Solidarity springs from shared humanity…AI governance should transform technological interdependence into consciously built solidarity directed towards the common good of the entire human family.” Absolutely…and that’s precisely what AI does, ‘governance’ notwithstanding.

Leo’s value of Subsidiarity “means a shift away from both state centralism and market self-regulation towards a distributed responsibility among various actors… Individuals and local communities should retain their agency… multi-stakeholder and public-private models…do not guarantee justice, accountability, or genuine participation.” Right.

AI Governance emphasizes two of Leo’s values above all: The universal destination of all goods and reliance on subsidiarity as the cornerstone of any regulatory framework. Amid Leo’s tangled jungle, our authors have identified two prize winning rose bushes.

But alas, our authors back slide: “The state in this model…remains the guarantor of rights, enforceability, and minimum safety standards.” Remains? The guarantor? We believe that rights are derived from human nature (image and likeness of God) per se, not from the generosity of the State (Declaration of Independence). And a survey of history reveals that states are more likely to infringe those rights than protect them.

“AI governance requires not only the protection of citizens (largely unnecessary) but also their active participation (very necessary). Given the limited agency of individuals in isolation, collective forms of organization and participation are essential. Their role is to expose power imbalances, represent vulnerable groups…and introduce into debate questions that neither the state nor the market asks.” Indeed.

All of which brings us back to the consensus goal (above): “A more just and inclusive social order in which technology enables human flourishing.” This is precisely the mission and promise of AI. It let us end wage slavery, fund a ‘living wage’ safety net, and reconstitute the human person as an end rather than a means. Of course, it needs conscious direction but most importantly, we need to let it do its job.

We need to prepare ourselves, politically and spiritually, for an era of Sustained Abundance; we must not cling to the familiar shores of scarcity, want, and wage slavery just because we fear to set sail on the uncharted waters that lead to universal prosperity. We must train ourselves to recognize and accept God’s grace in whatever form it takes.   

So can the Finns rescue Leo’s Papacy? I very much hope so…but sadly, I doubt it.