AI: an expression of a new (blessed) cognitive freedom

The factory that Leo XIII saw before him in 1891 was a complex comprising workshops, steam engines, looms, workbenches and workers. Today, a significant part of production is hidden from view: data centres, GPUs, and models trained far from the places where they are used. It is within these new infrastructures that the res novae take shape, which Magnifica Humanitas recognises in the ‘major trends’ of digitalisation, AI and robotics (MH no. 4).

Leo XIV’s encyclical addresses several issues: the nature of technology, algorithmic decision-making, automated warfare and the global order. One theme, more than any other, links it to Rerum novarum: the person in relation to work amidst technological transformations. It is in work, where the person “develops many dimensions of their existence” (MH no. 148), that an “essential key” to the social question lies (Laborem exercens, no. 3). ‘The human person always remains “the way of the Church”’ (MH no. 50). From this perspective, private property is legitimate but not absolute; it must be ordered in accordance with the social function and universal destination of goods, including ‘intangible and cultural goods’, ‘patents, algorithms, digital platforms, technological infrastructure and data’ (MH nos. 65–67).

AI infrastructures, the res novae of our time, are also the sites of a new concentration of economic power, involving “private actors, often transnational, endowed with resources and capacity for intervention superior to those of many governments” (MH no. 5). And, as a “decisive factor”, control over platforms, infrastructure, data and computing power lies in the hands of “major economic and technological players” who “set the conditions for access, the rules governing visibility and the very possibilities for participation” (MH no. 95): in large language models, proprietary weights and alignment criteria, with the risk of codes of ethics being defined by a small group of private actors; in computational capital, an extremely restricted supply chain for GPU production and a limited number of large-scale data centres; in programming standards and model libraries, difficulties in participation. The encyclical calls for digital subsidiarity, comprising independent audits, transparency regarding algorithms, equitable access to data, and means of redress (MH nos. 107–109 and 71).

Artificial intelligence is not merely a tool; “when it is used as a criterion, it ends up determining what matters and what can be discarded, reducing creation to an object of exploitation and people to cogs in a system designed to perform ever more efficiently” (MH no. 92). This risk is exacerbated by a distinctive feature of AI: “Modern artificial intelligences are more ‘cultivated’ than ‘built’”: “developers do not directly design every detail, but rather create an architecture on which the AI ‘grows’”, whilst “fundamental scientific aspects […] remain unknown at present” (MH no. 98). This had already happened with heat engines, which were used for decades before Carnot and Clausius formulated the theory behind them. For generative models, the key issues are those of interpretability and accountability: who is responsible when an algorithm selects, excludes or decides? The encyclical therefore calls for automated decisions to remain comprehensible and open to challenge, and for the chain of responsibility—covering those who design, train, authorise and deploy the systems—to be identifiable (MH nos. 164 and 199). This contrast is reinterpreted by the encyclical through the distance between Babel and Nehemiah: The former demands a ‘single language, including a digital one, capable of translating everything—even the mystery of the person—into data and performance metrics’ (MH no. 10): the hubris of a form of knowledge that seeks to reduce reality to a single code of measurement and performance. The latter entrusts each person with a section of the wall: the recognition of limits and shared labour set the reconstruction in motion once more.

When considering the transformation currently underway, the encyclical highlights its risks: concentration of power, exclusion, exploitation, and the adaptation of labour to machines. This thus revives an age-old concern, one that concerns the relationship between science, technology, capital and labour – from the passage on machines in the Grundrisse to ‘dead labour’ in the first volume of Capital: between a freedom possible only outside of labour and the subsumption of judgement within the machine and within proprietary systems. Faced with this alternative, *Magnifica Humanitas* points the way towards the protection of the person and emphasises how labour and time, rather than being liberated, must be directed towards the fulfilment of the human person. Technical progress is not at odds with the person (MH no. 4): it is ‘valuable in itself’ (MH no. 94), and technological innovation can be ‘a human form of participation in the divine act of creation’ (MH no. 111).

In history, that potential can only unfold in concrete realisations, always finite, always imperfect: new powers alongside new freedoms, new dependencies but also new spaces for initiative and expression at work, side by side. On the other hand, technological transformations—from the steam engine to electricity, from chemistry to computing—have given rise to new capabilities, new enterprises, new trades, and spaces in which individuals have been able to express themselves through their work. In Europe, science, technology and institutions have coexisted for centuries, supporting one another. Joel Mokyr identified the origins of this in the Republic of Letters, the European network which, from the seventeenth century onwards, circulated useful knowledge across languages and borders, fuelling the beliefs and institutions that made knowledge contestable and cumulative, and shaped a culture of growth.

Within that narrative lies an even older polarity. Magnifica Humanitas recognises it in Augustine’s two cities, the ‘struggle between two loves’ in human history (MH no. 130). From patristic theology right through to Popper, this dialectic has animated European thought, holding together otherwise irreconcilable positions and, through ruptures, conflicts and contradictions, settling into the balances from which universities, civic self-governance, rights and intermediary bodies have emerged. It is upon that mutual limitation, and upon a secularisation that first denied and then, in part, recognised its Christian origins, that inclusive institutions and mechanisms for limiting power have taken shape: the rule of law, independent authorities, competitive markets. Those balances are not a given and may be lost. In the new digital space, these very mechanisms demand to be rethought and put back into practice.

Magnifica Humanitas focuses on the question of personal fulfilment in the age of artificial intelligence. It is the question which, from Rerum novarum through Laborem exercens to Caritas in veritate, has run through over a century of social teaching. From the steam engine to the GPU, the material foundations of production have changed, and with them the object of technology. Artificial intelligence, within an open society that must be ceaselessly rebuilt, can become an instrument of a new cognitive freedom, which broadens the judgement of those who work. In work, through work.