Promoting xenomorphic patient-facing AIs: The case against anthropomorphism in medical AIs

The rapid emergence of patient-facing medical artificial intelligence (MAI) raises pressing questions about its design and impact on healthcare. Current anthropomorphic design strategies, which endow AIs with human-like features, are based on a reductivist understanding of healthcare and risks undermining the integrity of doctor-patient relationships (DPRs) that are foundational to positive health outcomes. This paper argues for a xenomorphic approach-designing MAIs with decidedly non-human, even alien-like characteristics. By distinguishing AIs from human physicians as far as possible, xenomorphic design may avoid direct competition with human relational roles, protecting the ontological distinctiveness of DPRs. Importantly, we suggest that such designs may foster a new class of therapeutic relationships with patients: non-human but nonetheless health-affirming. Analogous to the established benefits of animal-assisted therapies, xenomorphic MAIs could provide complementary sources of trust, comfort, and support without supplanting essential human care. The xenomorphic approach may therefore not only avoid the conceptual and practical challenges of anthropomorphism but also offer an innovative path to expand therapeutic possibilities and improve health outcomes.