JMIR Publications

ABSTRACT

Background:

Artificial intelligence (AI) has been increasingly used in healthcare settings, especially in the field of primary care. However, little is known about patients’ perceptions, beliefs, and attitudes towards this relatively new entity. We conducted a cross-sectional survey study on adult primary care patients in a single primary care clinic in Western New York to assess their perspectives on AI.

Objective:

To assess patient perspectives on artificial intelligence in primary care.

Methods:

A cross-sectional survey was conducted on adult patients in a private practice-based primary care clinic in Western New York between June 2025 and August 2025. Patients (>18 years old) were asked to complete an anonymous survey questionnaire during their clinic visit. The survey asked about their demographics, familiarity with AI, perceived benefits and concerns, and preferences regarding AI integration into their healthcare visits with their primary care provider. Of 767 respondents, 702 provided informed consent to participate. Descriptive statistics and subgroup analyses by age, race/ethnicity, and education were conducted. Study design was reviewed by an independent IRB and given exempt status.

Results:

Overall familiarity with AI in healthcare was mixed . 41% reported being somewhat or very familiar, 38% were unfamiliar or uncertain, and only 20% were very familiar. 60% of participants preferred that physicians use AI tools but retain final decision-making authority. Only 3% supported unsupervised AI decisions, and 37% preferred no AI involvement. On a 3-point Likert scale (1 = Not at all comfortable, 3 = Very comfortable), for AI assisting in diagnosis mean comfort ratings were 2.04,for suggesting treatment options 2.09, for chronic condition monitoring 2.06 , 2.01 for interpreting medical imaging, and 1.80 for AI-based chatbots. Most supportive of physician-reviewed AI (63.8%) were Younger respondents (18–39 years) , while older respondents (≥80 years) favored no AI involvement (53.6%). Major concerns reflected a strong desire to preserve human elements of care and ensure safety. The most frequently cited issues were lack of human interaction (74%), potential for medical errors (64%), misunderstanding of patient values (58%), job loss among healthcare workers (54%), and privacy or data security (52%). Only 7% reported no concerns. Openness to learning more about AI was evenly split (50.3% Yes; 49.7% No).

Conclusions:

Overall, patients expressed a cautious view of AI in healthcare. Some patients recognize AI’s potential to support diagnosis and treatment, however, the majority of the patients were reluctant on relying on AI for medical decision making. Main concerns were about lack of human interaction, data privacy, and the potential for errors, contributing to a general reservation towards widespread acceptance. Older age and lower education showed more mistrust in AI, highlighting the need for targeted education and reassurance in the future. Clinical Trial: Not Applicable.