In a recent study led by Duke Health, patients were surveyed to determine their preference for messages written by artificial intelligence (AI) versus human clinicians. Surprisingly, the results showed that patients actually preferred responses drafted by AI over those written by humans. However, this preference was slightly diminished when patients were informed that AI technology was involved in generating the responses.
Published in JAMA Network Open on March 11, the study revealed that overall satisfaction with communications, whether from AI or human authors, was high among patients. This suggests that patients do not lose confidence in the message even when they are made aware that AI technology was used.
Senior author of the study, Dr. Anand Chowdhury, expressed the importance of transparency in disclosing the use of AI in healthcare communications. The study aimed to measure the impact of disclosing AI technology on patient satisfaction and confidence in the messages they receive.
The survey was conducted among members of the Duke University Health System patient advisory committee, comprising Duke Health patients and community members. More than 1,400 individuals responded to the surveys, which focused on three clinical topics of varying seriousness.
Responses were provided by a team of physicians for the human-authored messages, while generative AI responses were created using ChatGPT technology. Participants were asked to rate their satisfaction, usefulness, and sense of care based on the responses they received.
Results showed that patients preferred AI-drafted messages over human-authored ones, with AI messages being perceived as more detailed and empathetic. However, when participants were informed that AI was involved in drafting the responses, satisfaction slightly decreased.
Despite this slight decrease in satisfaction, patients were still more satisfied with messages when they were not told AI was involved in the process. This finding highlights the importance of balancing the use of AI technology in healthcare communications with maintaining patient satisfaction and trust.
Dr. Joanna S. Cavalier, the first author of the study, emphasized the potential benefits of using AI technology to enhance communication with patients while addressing clinician burnout. The study’s findings provide confidence in the use of AI tools to improve healthcare communication while ensuring transparency with patients.
In addition to Dr. Chowdhury and Dr. Cavalier, the study’s authors include a multidisciplinary team of physicians and researchers. The study, titled “Drafted Responses to Electronic Messages,” offers valuable insights into the impact of AI technology on patient satisfaction and communication in healthcare.
For more information on the study, readers can access the full article in JAMA Network Open with the DOI: 10.1001/jamanetworkopen.2025.0449. The research was conducted at Duke University Medical Center, a leading institution in healthcare research and innovation.
Overall, the study sheds light on the evolving role of AI technology in healthcare communication and the importance of maintaining patient satisfaction and trust in the process. As healthcare systems continue to integrate AI tools, transparency and patient-centered care remain crucial considerations for improving patient outcomes and experiences.
