AI and healthcare: some considerations on trust

Abstract

This article explores the relational nature of trust in artificial intelligence (AI) systems applied to the medical field. In contrast to approaches that tend to locate trust in individual properties—either of the trusting subject or of the object of trust—we propose understanding it as an emergent phenomenon that can only unfold fully within the interaction between humans and technologies. We critically analyze several key concepts in the current debate—such as responsibility, accountability, anthropomorphism, and value alignment—showing that none of them can be unequivocally attributed to a single pole of the relationship. We argue that an adequate understanding of these constructs requires situating them within a relational perspective, where trust does not simply derive from technical qualities or subjective attitudes, but from shared structures of meaning, practices of co-responsibility, and appropriate institutional frameworks. This approach allows for a more precise engagement with the ethical challenges of medical AI and guides the design of systems that are not only efficient but also trustworthy in a robust sense.