Gastroenterology (GI) is a leading field for AI innovation, assisting clinicians with dynamic endoscopic imaging and procedures. But what do patients think of this growing component of care?

To answer that question, researchers led by gastroenterologist Vinay Jahagirdar, MD, of the Division of Gastroenterology, Hepatology & Nutrition at Virginia Commonwealth University in Richmond, Virginia, conducted a multicenter, survey-based study of patient perspectives in the US and Ontario, Canada’s most populous province.

Reported in the American Journal of Gastroenterology, the results of the group’s 25-item questionnaire showed that while GI patients generally support AI as an assistive tool, they prioritize physician oversight and transparency, with an overwhelming majority (93%) saying the physician should retain final decision-making authority. Knowledge of and trust in AI varied appreciably by demographic factors such as age, sex, region, employment status, and levels of education and income.

The investigators concluded that addressing AI knowledge gaps and demographic disparities is essential for the equitable adoption of this tool in clinical GI.

From the clinician’s standpoint, GI is one of the most AI-ready specialties. “So much of what we do is image- or video-based — endoscopy, capsule endoscopy, pathology — and much is also pattern-recognition-heavy,” Jahagirdar told Medscape Medical News. “AI can augment detection and classification, as in polyps or Barrett esophagus, and standardize quality, especially in high-volume workflows.”

That said, however, physician knowledge of patient attitudes lags behind developments experienced by clinicians. The cross-sectional survey of 230 adult GI outpatients (59% female; mean age, 50 years) found that many had only low-to-moderate baseline familiarity, but most were still receptive to AI as a supportive tool when paired with clinician oversight.

Among the survey’s specific findings:

• A majority reported either very little (28%) or a little (31%) familiarity with AI, while 31% reported moderate knowledge, and only 9% reported having significant knowledge.
• Greater AI knowledge was associated with male sex, younger age, employed status, and higher education and income levels.
• Trust in AI was moderate: 30% expressed “quite a bit” or “strong” trust, with 60% expressing concern about AI’s reliability in clinical decision-making.
• By region, trust levels varied significantly, with respondents from the US West Coast and southern Canada reporting higher trust and those from the US Midwest and Southwest reporting comparatively lower trust.
• Employment status also affected trust, with full- and part-time workers expressing more confidence than retired or unemployed individuals.
• Trust in endoscopic AI was significantly greater among participants with postgraduate education.

In broaching the issue of AI with patients, Jahagirdar advises gastroenterologists to use plain language emphasizing that AI is a decision support tool, not a replacement. “Explain that it may highlight findings or provide a second set of eyes, but the physician makes the final decision,” he said. Patients also want explicit disclosure when AI is part of care and a brief explanation of benefits, limits, and the way data are protected.

“Reassurance and clarity matter: About 74% of respondents stressed the importance of being informed of AI use,” he said.

What about addressing AI hesitancy or outright refusal? While specific refusal behavior was not assessed in the study, Jahagirdar said, “What we can say is that hesitancy is tied to concerns of reliability, privacy, and cost, so if patients object, it is usually addressable with transparency, data-security explanation, and clear clinician accountability.”

Clinically, the implementation signal from the survey is clear. Patients are broadly open to AI if it stays physician-led and is disclosed and explained. The practical steps are simple, said Jahagirdar. “Standardize an AI disclosure line in consent/workflows, emphasize clinician responsibility, and provide brief patient-facing education to support equitable adoption across different ages, education levels, and trust baselines.”

Offering a nonparticipant’s perspective on the research, Cadman L. Leggett, MD, gastroenterologist at the Mayo Clinic in Rochester, Minnesota, said the survey provides a useful overview of patient concerns. He noted that patients at his center seem quite familiar with AI and understand that it can improve diagnostic performance.

“They expect us to use AI algorithms along with other technologies. It’s just part of technology in medicine, and I haven’t noticed much skepticism or hesitancy,” he told Medscape Medical News.

Leggett agreed that patient fears center mainly on data privacy and information hacking. “But our patients are aware that a lot goes into de-identifying and safeguarding their private information,” he said. If patients express concern, “That’s a strong reassuring message to send.”

This research received no specific funding. The authors and Leggett reported having no conflicts of interest to disclose.