Answers are grounded in this paper and cite their sources.
Objectives Health systems are rapidly expanding their use of artificial intelligence (AI) across the care continuum. However, best practices for explaining AI to patients to build trust and support positive experiences remain unclear. Methods We conducted three waves of interviews with adult patients across a large health system, with each phase building upon prior findings. Phase 1 (N = 5) explored expectations for responsible AI use and communication preferences, which helped inform development of the system’s AI evaluation framework. Phase 2 (N = 5) assessed patients’ informational needs following framework implementation and guided creation of draft patient-facing materials. Phase 3 (N = 8) used cognitive walkthroughs to evaluate these materials, which covered the system’s approach to responsible AI and solution-specific applications. Transcripts were analyzed using thematic analysis and inductive and deductive coding. Results Across phases, four themes emerged. Patients wanted (1) clear explanations of how AI supports clinical decision-making without replacing clinicians, (2) concrete examples of AI tasks, (3) clear rationales for adoption, and (4) transparent information about safety, privacy, and data protections. Phase 3 yielded actionable guidance for improving communications, including reducing technical or surveillance-oriented language, simplifying descriptions of AI functions, and centering patient safety and experience. Patients emphasized the need for straightforward explanations of data use and whether AI-supported care is optional, and the availability of plain-language, visually supported materials tailored to diverse literacy and language needs. Conclusions Patients value clear and transparent communication about how AI supports care, particularly regarding safety, data use, and patient choice. Our findings offer practical guidance for developing patient-centered AI communication. Public interest summary Artificial intelligence (AI) is becoming increasingly common in healthcare, but many patients are unsure and/or have concerns about what it means for their care. In this study, we interviewed patients from across a large health system to learn what information they want when AI tools are used by their care teams. Patients wanted simple explanations of what AI tools do, examples of how they are used, clear reasons for why they were adopted, and details about how AI supports their doctors and nurses but does not replace them. Trust also depended on patients knowing how their data are protected, who oversees the AI tools, and whether AI-supported care was optional. Patients highlighted the importance of straightforward materials that avoid jargon, offer visual aids, and are accessible to people of different backgrounds and abilities. These findings may help health systems explain AI in ways that improve patient understanding and trust.
No takes yet. Share an insight, caveat, or question.
Kramer et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: