Chaplains' Reflections on the Design and Usage of AI for Conversational Care

Non-clinical emotional support requires different AI design than clinical diagnosis: The paper found that most conversational AI development focuses on clinical expertise for diagnosis and intervention, but everyday emotional support—like grief counseling and personal reflection—needs fundamentally different conversational approaches that current chatbots struggle to provide.

Chaplains identified four critical gaps in AI chatbots: Through analysis of eighteen chaplains' experiences building chatbots, the research identified specific limitations across Listening (understanding nuance), Connecting (building genuine relationships), Carrying (bearing emotional weight with someone), and Wanting (understanding what people truly need)—suggesting current AI lacks the relational depth required for meaningful care.

"Attunement" should guide responsible AI design for care: The paper proposes that AI for well-being should be designed through a relational lens of "attunement"—the ability to understand and respond to delicate human experiences—rather than focusing primarily on clinical outcomes or problem-solving.

Domain expertise beyond medicine is essential for responsible AI: The research demonstrates that responsible AI development requires input from diverse professionals like chaplains, not just clinicians, to ensure conversational AI serves the full spectrum of human needs including spiritual and emotional support in everyday contexts.

Most chaplains expressed cautious skepticism about chatbot limitations: While some viewed AI chatbots optimistically, the majority identified significant limitations in their ability to authentically support everyday well-being, suggesting current technology may not be ready for independent deployment in pastoral care contexts.