Diego Rimer & Rubén Sulbarán
20 September 2026
Note on the use of artificial intelligence: The authors used artificial intelligence tools to support their research, drafting, and editorial refinement. They reviewed and verified the materials and take full responsibility for the arguments, sources, and final text.
In this age of “super productivity,” accelerated by artificial intelligence (“AI”), one question has slipped off the agenda: what is lost when efficiency becomes the sole criterion by which decisions are made? The question is not nostalgic. It goes to how a society chooses what to preserve as it advances.
Pope Leo XIV’s first encyclical, Magnifica Humanitas: On Safeguarding the Human Person in the Time of Artificial Intelligence, offers the sharpest recent formulation of that concern.1 We take it here not as a premise to be accepted on faith, but as one of the clearest available illustrations of a problem that reaches every reader, believer or not: a technology optimised for output is quietly reshaping the grounds on which human critical judgment stands.
The concern is not new. A decade ago, Pope Francis warned in Laudato Si’ against what he called the technocratic paradigm as: “the tendency to let the logic of efficiency, control, and profit alone shape personal, social, and economic decisions.”2 Where Francis named the paradigm, Leo XIV applies it directly to digital power and to the systems that now mediate work, information, and decision-making.3 The continuity matters: this is not a reaction against one technology, but a sustained argument considering that efficiency, left unchecked, becomes a way of seeing the world, and eventually a way of seeing people.
Here lies the heart of the matter. AI has become mainstream not merely because it marks another technological threshold, but because it promises to reduce and even remove human interaction from most processes, and, with it, the friction of human error. Yet that friction is also where moral judgment lives. Removing the human from the loop does not simply make a process faster; it removes the point at which someone must weigh, hesitate, and take responsibility. The common rejoinder — that AI will handle the “non-critical” decisions while humans retain the “key” ones collapses on inspection.
As Leo XIV puts it, AI cannot be treated as morally neutral, even as the tool is described and “developed” as amoral: every tool embodies choices through what it measures, ignores, and optimises, and how it classifies people and situations; a system that treats some lives as less worthy, or excludes them without appeal, has already introduced criteria that contradict human dignity.4 And there is no decision — not even the ordinary ones we make daily that operates without some element of priority or conscience. If that is so, then there is at present no framework ensuring that any moral consideration enters either the critical or the non-critical decision. The risk Francis named remains exact: “an increase in means without a growth in humanity — ‘having more’ without ‘being more.’”5
What gives this unusual force concern is that it is now being voiced from inside the industry itself. In September 2026, Jacob Coxon resigned after three years of pretraining research at OpenAI and Anthropic, warning that competitive pressure was driving frontier laboratories toward self-improving systems faster than anyone could make them safe.6 Days later, Anthropic’s own chief executive, Dario Amodei, called publicly for the industry to slow the pace of frontier development so that safety research and independent oversight could keep up.7 He grounded the appeal in concrete developments that are considered key safety concerns. Within hours, even professional and commercial adversaries, Sam Altman and Elon Musk, among others endorsed the direction of his argument.8 The people closest to building these systems are beginning to doubt, out loud, that efficiency and competitive advantage should remain the overriding goals.
The obvious objection is that morality in decision-making is unauditable and unproductive, that, to demand it is simply to demand a brake on progress. But that framing is exactly the one the industry’s own leaders have now rejected. Amodei is careful not to argue for stopping, because not building the technology deprives humanity of its benefits, or leaves it to authoritarian powers, while building it too fast is reckless. The choice is not between progress and restraint, but about pacing preserving the time and the human vantage point from which risks can still be seen and governed. That answer does not come from a pulpit. It comes from within the race itself, which is precisely why it is hard to dismiss as moralism.
We are witnessing a level of productivity unlike anything seen before. Yet as human input is steadily replaced by automation, we edge closer to a point where human contribution itself begins to feel out of place across many industries. Pope Leo XIV puts it plainly:
“When efficiency becomes the ultimate measure of value, human beings are tempted to see themselves as a project to be optimised rather than as persons called to relationship and communion.”
Here, “communion” need not be understood as a profession of Christian faith; it can be read more broadly as the human capacity for relationship, mutual responsibility, and life in common. His words invite a debate we can’t afford to skip: is efficiency the only thing AI should strive for? And is efficiency at any cost what humanity really needs?
About the authors
Diego Rimer: Corporate legal counsel; AI Regulatory Strategist; LL.M. in European and International Law; Ph.D. candidate in Artificial Intelligence Law. Luxembourg, Grand Duchy of Luxembourg.
Rubén Sulbarán: Legal counsel; Service Delivery Manager; Professional Scrum Master I. Montevideo, Uruguay.
Notes
- Leo XIV, Magnifica Humanitas (Encyclical Letter, 15 May 2026) vatican.va (Last accessed 16 September 2026). ↩
- Francis, Laudato Si’: On Care for Our Common Home (Encyclical Letter, 24 May 2015) para 107 vatican.va (Last accessed 16 September 2026). ↩
- Leo XIV, Magnifica Humanitas (n 1) para 92. ↩
- ibid para 104. ↩
- Francis, ‘Address to Participants in the Meeting for the “Council for Inclusive Capitalism”’ (Vatican, 11 November 2019) vatican.va (Last accessed 16 September 2026). ↩
- Cf. Harry Booth, ‘He Helped Build Powerful AI at OpenAI and Anthropic. Now He’s Afraid It Could Kill Us’ TIME (9 September 2026) time.com (Last accessed 16 September 2026). ↩
- Dario Amodei, ‘We Must Pace the Frontier’ (darioamodei.com, 12 September 2026) darioamodei.com (Last accessed 16 September 2026). ↩
- ‘Anthropic’s Amodei Calls for AI Oversight, Joined by Altman and Musk’ The Washington Post (12 September 2026) washingtonpost.com (Last accessed 16 September 2026). ↩