Artificial intelligence and medicine: an epistemological inquiry

Artificial intelligence (AI) in medicine inspires both enthusiasm and concern. It introduces an unprecedented development in the history of knowledge: for the first time, understanding is being generated by a non-human agent. This shift takes on particular weight in medicine, where the ultimate subject is the human being. AI thus invites a deeper epistemological reflection—on both the scientific structure of medical knowledge and its philosophical underpinnings. The rise of AI prompts the question, framed by Thomas Kuhn, of whether we are witnessing a paradigm shift in medical reasoning, traditionally anchored in evidence-based medicine. Two major epistemological disruptions emerge. The first is the opacity of AI systems—the ‘black box’. The second is the appearance of non-human knowledge, whose logic may elude traditional frameworks. This article explores these challenges through the lens of Immanuel Kant’s three questions: ‘What can I know?’ ‘What ought I to do?’ ‘What may I hope?’ At stake is the ability to build feedback mechanisms for machine-generated knowledge, restore intelligibility to opaque processes and avoid the flattening effect of epistemic monoculture. The challenge is not only epistemological but also ethical. In the final phase of his thought, Kant added a fourth question: ‘What is the human being?’ In the age of AI, that question gains new urgency.