How did patients and doctors rate the conversations?

Patients appeared receptive to the technology. They rated the quality of AMIE's conversations favorably—as did physician evaluators—across most measures. Patient attitudes toward AI improved after interacting with the chatbot and remained elevated after their appointment with a clinician.

"One of the most important questions was simply how patients would engage with a conversational AI system in a real clinical setting," said co-first author Jacob Koshy, M.D., M.P.H., a hospitalist in the Division of General Medicine at BIDMC. "By prospectively observing these interactions, we were able to better understand both the opportunities and the limitations of the technology in the context of routine patient care."

Where did trust fall short?

Patients gave high marks to the system's ability to listen, explain information and help them feel at ease. But two concerns persisted:

• Trust in the confidentiality of information shared with the system
• Trust in the chatbot's honesty and overall trustworthiness

The researchers said these findings underscore the importance of continued work to build patient trust as AI tools move into clinical care. "Research suggests that patient trust in medical AI systems flows from relationships with their human providers," Rodman said. "Future research will need to explore which interaction characteristics can build patient trust and how AI interactions can enhance the patient-physician relationship."

Did the AI help doctors prepare for visits?

The findings also point to possible practical value in clinical workflows. Primary care providers were able to review an AI-generated transcript or summary before seeing their patients in 44 cases. The reported results:

• In 75% of cases, physicians said reviewing the material helped them prepare for the visit
• In 57% of cases, they reported it may have influenced their clinical approach
• In one case, a provider called the interaction somewhat harmful, citing concern that a patient may have experienced anxiety after AMIE included lymphoma among its possible diagnoses

What the study does not show

Rodman and colleagues emphasized that the study was designed to evaluate feasibility—not to determine whether AI improves health outcomes. That question remains open, and the small sample of 98 completed encounters at a single medical center limits how far the results generalize.

"This study helps establish the baseline characteristics of such real-world conversations," said co-first author Peter Brodeur, M.D., a clinical fellow in cardiovascular medicine at BIDMC. "These baseline characteristics provide a foundation for all future prospective trials aimed at discovering when human-on-the-loop workflows may be appropriate."

Co-senior author Marc L. Cohen, M.D., clinical chief of primary care in the Division of General Medicine at BIDMC, framed the study as a template for how AI safety might be assessed before clinical implementation. "We believe the future involves continued involvement of human clinicians in patient care, but see a future where the triad of patient, human clinician and AI helps to provide our patients with the very best evidence-based care possible," Cohen said.

The paper is published as Peter G. Brodeur et al., "Conversational diagnostic artificial intelligence in ambulatory primary care: a prospective feasibility study," The Lancet (2026). DOI: 10.1016/s0140-6736(26)01535-7.