AI may be a tool of care, but it can never be a bearer of care

Christina Koenes

MD and master Philosophy
General practitioner, Munich

Lecture given during the conference “AI and medicine: the challenge of human dignity”, organised by Pontifical Academy for Life and International Federation of Catholic MedicalAssociations (FIAMC), November 10-12-2025, Rome

Artificial Intelligence (AI) has recently captured enormous attention, both in academic discourse as well as in the public sphere. Yet the assessments of AI could hardly be more polarized: for some, it appears as an uncontrollable force which threatens to dominate or even abolish humanity; for others, it represents a universal remedy – a harbinger of progress that promises to overcome the limitations of the human condition. 

At its core, Artificial Intelligence can be understood as the capacity of computers to discern patterns and relationships within vast datasets — patterns that would otherwise remain hidden to the human mind due to their sheer scale. Through remarkable speed and precision, AI systems simulate what is commonly called “intelligent” behaviour. This capability offers tremendous advantages: accelerating processes, enhancing efficiency, and minimizing errors. However, the critical question arises: can we simply deploy such systems without limitation —without reflection on the values and ethical frameworks that ought to guide their use? 

Collingridge dilemma 

The Collingridge dilemma reminds us that technological control presents a paradox: As long as a technological invention is still easily controllable, it is poorly understood; yet once its long-term effects become evident, it has already become deeply embedded and exceedingly difficult to regulate (1). With AI, we may currently find ourselves in a transitional stage between these two poles. At the same time both too little and too much is being understood: we witness tangible impacts in daily life, yet our ability to foresee the deeper consequences remains limited. Precisely for that reason, now is the right moment to steer its development—to ensure that the values and worldview underlying AI are grounded in respect for human dignity and ordered toward the common good: the flourishing of each individual and all of humanity. 

How AI diverges from human intelligence in 3 dimensions

Three dimensions are particularly crucial in understanding where AI diverges most profoundly from human intelligence. 

1. The Capacity for Truth and Morality 

Since Plato, truth has been most clearly defined by the adequation theory: veritas est adaequatio rei et intellectus—truth as the correspondence between reality and intellect (2). Knowledge therefore mediates between the knowing subject and the world; it exists “in-between”— metaxy in Greek —between being itself and human understanding (3). 

AI, however, is inherently self-referential. It processes information within a closed system, lacking any genuine relation to external reality. That means: AI processes data drawn from the world, yet it has no relationship to the world itself. Consequently, the metaxy—the bridge between knowledge and being—is absent. Hence, AI is not capable of truth in the philosophical sense. This does not mean that its responses are necessarily false—its data processing often yields remarkably accurate outputs, but it lacks the ability to verify truth from within itself. 

AI processes information within a closed system, lacking any genuine relation to external reality. 

Equally significant is the fact that AI operates according to statistical regularities.
What occurs most frequently—what falls within the “normal” range of a Gaussian distribution (4)—is deemed correct. But the human being is, by nature, the exception. There is no “average” human. If ninety-nine out of a hundred people suffer toothache, no one would pity the one who does not—though he is the “statistical anomaly.” Rather, we would seek to cure the ninety-nine, as being without pain is intrinsically valuable (5). Moral reality functions in a similar way: something is not good because it is common; it is good because it has intrinsic value. 

AI, confined to statistical correlation, cannot make value judgments. Out of AI´s perspective, nothing carries intrinsic meaning or worth. While statistical normality may coincide with truth or goodness by chance, this occurs only when the data it processes is itself grounded in sound human judgment. Therefore, AI possesses neither the capacity for truth nor the capacity for moral discernment. These remain distinctly human faculties. 

2. Embodiment 

The human being is an inseparable unity of body and soul—a living whole, not merely a sum of parts. As the philosopher Helmuth Plessner expressed, a person has a material body but is a living body. (6) Through embodiment, we are capable of being touched, affected, and, in turn, of touching and affecting others. (7) To touch, in the deepest sense, means to leave an imprint—to carry within oneself the impression of others, and to leave one’s own impression upon them. This occurs prior to any deliberate act of cognition. We do not need to know someone’s biography to be moved by them.  

Pain reveals this vividly—whether physical, psychological, or emotional. Despite all attempts to numb it, we always feel that initial pain which reminds us that we are alive, open to the world and to others. Our capacity to be affected and to affect lies at the very heart of our humanity. 

AI, by contrast, is a merely material system—albeit one that can imitate conscientiousness with astonishing fidelity. Yet it can neither truly touch nor be touched.
All external inputs are instantly absorbed and assimilated into its internal processes; there is no “other” encountered as other. AI remains a closed circuit, incapable of dwelling in relationship. 

3. Personhood 

To be a person is to be an “I”—a subject—and, for others, a “me”—an object of equal standing. Being an “I” includes both selfhood—the answer to “Who am I?” — and self-distance — the capacity to reflect upon oneself, to become an observer of one’s own being (8). Within this lies the interconnection of being and becoming: the I in the present, shaped by the past, always oriented toward the future. This dynamic enables growth, creativity, and self-transcendence—the realization of one’s potential—possum: “I can be.” 

These two capacities—selfhood and self-distance—allow each person to occupy a unique position in space and time, from which only they can perceive, encounter, and act. 

Genuine encounter requires such self-distance; it is the foundation of moral behaviour. It enables us to recognize the other as a person and prevents the self from collapsing into itself. In responding to another’s existence, we acknowledge them as “you” standing before “me”. Thus, the I becomes an I through the encounter with the you (9). 

Thus, only the human being is capable of forming reciprocal relationships—and therefore of truly encountering the other 

AI lacks this dimension entirely. It cannot answer the question of its own being, nor can it take distance from itself. It has no locus in the world, no interiority from which to encounter another. It cannot assume an excentric perspective. It exists only as a moment of data processing—not as a being that has become or is becoming. It has no past, no present, and no future. AI cannot claim, nor can it inhabit, the relational space of personhood. It remains a sum of its parts. 

Thus, only the human being is capable of forming reciprocal relationships—and therefore of truly encountering the other. We are created in the image and likeness of God, who is, in Himself, relationship—in the Trinity—and whom we, as human beings, reflect.

Especially as physicians, who care for the most vulnerable among us, it is crucial that we enter into these deep relationships. AI can never replace this. 

AI may be a tool of care, but it can never be a bearer of care. It needs to serve us in serving. 

Acknowledgement
The author would like to thank the Executive Committee of Societas Lucae.  

References

1. Collingridge, David: The social control of technology, St. Martin’s Press, New York, 1980 
2. Aquinus, Thomas:  Quaestiones disputatae de veritate. Über die Wahrheit. (German Edition), Marixverlag GmbH, Wiesbaden 2013. 
3. Plato: Politeia. Der Staat (German edition), Anaconda Verlag GmbH, Köln, 2019. 
4. Gauss, Carl Friedrich: Theoria Combinationis Observationum Erroribus Minimis Obnoxiae, Dietrich Verlag, Göttingen, 1823. 
5. Spaemann, Robert: Glück und Wohlwollen, Klett-Cotta Verlag, Stuttgart, 1989. 
6. Plessner, Helmuth: Lachen und Weinen, in: Philosophische Anthropologie, Frankfurt 1970. 
7. Henry, Michel: Das Wesen des In-Erscheinung-Tretens, Verlag Karl Alber, Freiburg/München, 2019. 
8. Plessner, Helmuth: Die Stufen des Organischen und der Mensch. Einleitung in die philosophische Anthropologie, Walter de Gruyter Verlag, Berlin/New York, 1975. 
9. Buber, Martin: Ich und Du, Philipp Reclam jun. Verlag, GmbH, Ditzingen, 1983