Patients prefer AI-written messages … until they learn “who” wrote them

Dr. Gregory House might have been proven right that working on his bedside manner was a waste of time—at least when it comes to electronic messages, patients apparently prefer hearing from an algorithm over flesh-and-blood physicians. Two new Stanford and Duke studies published in JAMA Network Open have found that when patients ask medical questions online, they’d rather read an AI-written communication, as long as they don’t know its origin.

The computer will see you now

When Stanford researchers compared responses to 59 clinical questions, survey participants consistently gave AI-generated answers higher marks (3.96 out of 5) than those written by actual clinicians (3.05 out of 5).

“We found that AI-generated responses were received more positively by patients, and that patients preferred and were more satisfied with the responses generated with the help of AI,” said Dr. Eleni Linos, director of Stanford Center for Digital Health in an interview. “That was surprising because many of us worry about doctors not paying attention to problems or not being there for us. The fact that patients themselves were more satisfied with these responses was surprising, but in a good way.”

AI-drafted for the win: Length matters

It turns out AI has one significant advantage over busy doctors: verbosity. While physicians dashed off responses averaging a mere 254 characters, AI systems took their sweet time with messages averaging 1,470 characters. The study found that for doctor responses, length significantly correlated with patient satisfaction—suggesting that in the world of electronic health records, patients equate wordiness with caring.

Don’t mention the “AI-written” part

A complementary Duke University study found that while patients preferred AI-drafted responses, they had slightly decreased satisfaction when told AI was involved. The Duke researchers surveyed 1,455 people about their preferences for electronic message responses, finding a mild preference for AI-written messages but higher satisfaction when they thought humans wrote the response.

“Although AI disclosure may slightly reduce satisfaction, disclosure should be maintained to uphold patient autonomy and empowerment,” the Duke researchers concluded. They found that when asked about preferred disclosure language, patients favored simple acknowledgments like, “This message was written by Dr. T. with the support of automated tools.”

This seemingly contradictory finding speaks to our complex relationship with technology—we want the benefits of “AI-written” text without necessarily knowing the source. Patients liked AI’s detailed, thorough responses better, but when told an algorithm wrote them, satisfaction dipped slightly.

Old guard meets new code

Not all physicians are ready to welcome their new AI overlords. “As a 78-year-old family doctor, recently retired after 50 years of practice, I am thankful, I think, that I will not have to incorporate artificial intelligence into the way I care for my patients,” commented Dr. George Benton III, who practiced medicine in rural North Carolina. “Print, text, or email communication with a patient can never, in my opinion, create the connection or capture the nuances of a real conversation.”

Another physician suggested researchers might be barking up the wrong technological tree: “Instead of training AI modules to imitate human communication, could we not use them to create better EHR interfaces, which to this day are painstakingly coded with variable success? This could pave the way for better integration of regional health care systems, as well as ease the constantly steep learning curve of each new EHR roll-out.”

A digital prescription for modern medicine?

Despite the grumbling, Dr. Linos remains cheerfully optimistic that AI could actually improve the doctor-patient relationship rather than replace it.

“It definitely gives me hope for the future because if we can create a win-win situation where doctors can get back some time for real in-person communication with their patients and the medical work they’re so good at, and patients are getting more in-depth, longer, more detailed, more accurate responses to their queries, that’s a really unique situation where this type of technology is helping both the patient and the clinical team, increasing efficiency, and no one loses,” she explained.

The vision is enticing: AI handles the lengthy explanations about minor concerns, freeing doctors to focus on face-to-face care. Meanwhile, patients get the detailed answers they crave without waiting days for a harried physician to type out a reply between appointments.

As healthcare moves further into the digital age, perhaps we’re discovering that sometimes patients just want someone—or something—to take the time to explain things thoroughly, even if that “someone” happens to be made of code. Just don’t remind them of that last part.