- In a small study, patients generally accepted AI-drafted messages for routine communication.
- Perceptions about message empathy, tone, and length varied according to circumstances and stakes associated with a communication.
- Transparency remains essential, as patients overwhelmingly demanded explicit disclosure of AI use to maintain trust in their healthcare relationships.
Patients care less about whether a message came from a human or artificial intelligence (AI) than the tone, length, and details, investigators in a small study concluded.
Patients expressed “high comfort” with AI-drafted portal messages, which did not influence preferences about message tone or expressions of empathy. Whether the messages seemed appropriate for the purpose and stakes held more sway.
The study participants broadly favored disclosure of AI use, calling transparency important for trust, reported Kellie Owens, PhD, of New York University Grossman School of Medicine in New York City, and colleagues in JAMA Network Open.
“Our study suggests that patients are generally comfortable with the use of generative AI to draft in-basket responses, particularly when framed as a workflow support that could improve response speed and reduce clinician burden,” the authors wrote. “Patients emphasized that the primary value of portal communication is timely resolution of health-related questions rather than personalization or relational depth.”
“This framing is imperative for health systems implementing AI-assisted messaging because it suggests that patients may evaluate these tools less in terms of whether AI replaces human connection between clinicians and patients and more in terms of whether AI improves reliability, clarity, and responsiveness,” they added.
Patient expectations about message tone and content changed with clinical context, Owens and team noted. For example, some patients preferred more detailed or emotionally supportive language in higher-stakes scenarios, such as interpretation of potentially concerning test results. That observation contrasted with findings from a previous survey, which showed that topic seriousness did not significantly affect patients’ messaging preferences.
Perceptions of Empathy
The study provided important insight into patient perceptions of empathy in healthcare communications and useful guidance for clinicians and health systems that implement AI into portal messaging, according to the authors of an invited commentary.
“Clinicians could more easily review succinct, AI-drafted text for the lower-stakes correspondence that regularly clogs clinicians’ in-baskets and crowds out time at the bedside,” wrote Teva Brender, MD, and Elizabeth Dzeng, MD, PhD, MPH, of the University of California San Francisco. “If unable to convey higher-stakes information in person or by phone, clinicians might use AI to deliver more effective, empathic electronic communication. Instead of dismissing AI’s remarkable capabilities … we should consider how AI might be leveraged, such as workshops, simulation with standardized patients, role-play, feedback, and coaching, to improve clinicians’ communication skills.”
“As one of the first qualitative investigations on the subject, the study by Owens et al. contributes essential nuance to the growing literature on AI-drafted portal messaging,” they added. “For now, while respondents in this study were comfortable with AI’s use, disclosure and oversight remain essential. It appears that patients are not ready for clinicians to turn their in-baskets over to machines.”
Clinicians and health systems have rapidly incorporated AI-generated drafts into patient messaging portals, which raises questions that extend beyond efficiency, Owens and colleagues stated in their introduction. Patient message portals have evolved into sites for information exchange, guidance, and interpersonal care. Patients may interpret message tone, length, and responsiveness as indicators of empathy, attentiveness, and professionalism.
In the researchers’ prior survey, participants generally preferred AI-drafted responses over clinician-written responses and rated them more satisfactorily across multiple messaging scenarios. However, satisfaction was somewhat lower when participants were explicitly informed that a response was AI generated.
“This paradox — preference for AI-drafted messages alongside reduced ratings when AI involvement is disclosed — suggests that patient evaluations are not only shaped by the content of messages but by beliefs about what AI use means in the context of clinical relationships and delivery of healthcare,” Owens and colleagues wrote.
Study Design, Key Findings
To continue exploring patient perceptions of AI-generated messages, the researchers recruited 40 adults who had participated in a prior survey about portal messaging. The objectives were to learn more about patient perceptions of AI-drafted responses in a messaging portal and identify implications for patient-centered implementation.
Women accounted for the majority of the participants (75%), 32.5% were white, 35% were Black or African-American, and a third were ages 65 or older.
The patients participated in videoconference interviews that lasted 45-60 minutes. The investigators used vignette-based prompts to explore patient perceptions about messaging associated with clinical scenarios, such as medication refill requests, side effects, and potentially abnormal test results. After a series of open-ended questions, patients evaluated messages associated with specific scenarios, which included examples of patient messages paired with draft responses that varied in tone, length, and implied authorship (AI or human).
Participants viewed portal messaging as useful for routine clinical needs but not for situations when they expected deep interpersonal connection. Most participants framed portal messaging as “fundamentally transactional,” suitable for resolving situations and obtaining routine information quickly. As a result, even “AI-hesitant” patients considered AI involvement in portal messaging more acceptable than use in direct decision-making or in-person interactions.
In general, patients were comfortable with AI-drafted portal messages, perceiving the technology as a reasonable response to clinician workload. However, the comfort level was conditional, as patients “repeatedly emphasized that AI drafting was acceptable so long as messages were reviewed and approved by a clinician before being sent, even for lower-stakes tasks.”
Perceptions about what constituted an “acceptable portal message” were heterogeneous, the authors noted. Preferences varied widely, shaped by clinical context and the perceived stakes of the circumstances associated with the message. Preferences for message tone and content also depended on the clinical stakes.
The patients expressed broad consensus on appropriate disclosure of AI use. All patients expressed a desire for transparency about AI use across healthcare contexts, not just portal messaging.
“Our data suggests that patients evaluate portal messages for signals of authenticity and care based on their conscious or unconscious expectations for how these concepts are expressed in digital communication,” the authors concluded. “Additional research would be helpful to understand the implicit values and criteria patients use to define empathy and care in health communication.”
