Attitudes of the Iranian Public Toward the Clinical Use of Artificial Intelligence in Medicine: A Cross-Sectional Survey

Mohammad Sina Maleki,1 Leyla Sahebi,2 Zahra Shahvari,3 Shahram Samadi4,5 1School of Medicine, Tehran University of Medical Sciences, Tehran, Iran; 2Maternal, Fetal and Neonatal Research Center, Family Health Research Institute, Tehran University of Medical Sciences, Tehran, Iran; 3Department of Nursing and Midwifery, Gachsaran Branch, Islamic Azad University, Gachsaran, Iran; 4Sleep Breathing Disorders Research Center, Imam Khomeini Hospital Complex, Tehran University of Medical Sciences, Tehran, Iran; 5Anesthesia, Critical Care and Pain Management Research Center, Tehran University of Medical Sciences, Tehran, IranCorrespondence: Shahram Samadi, Sleep Breathing Disorders Research Center, Imam Khomeini Hospital Complex, Tehran University of Medical Sciences, Keshavarz Blvd, Tehran, 1419733141, Iran, Tel +982161192053, Fax +982166581537, Email [email protected]: The integration of Artificial Intelligence (AI) into healthcare is a global trend, yet its success hinges on public acceptance. This study aimed to provide one of the first comprehensive assessments of awareness, trust, and preferences regarding the clinical use of medical AI among a sample of Iranian hospital visitors and to identify the key predictors shaping these views to inform national digital health policy.Patients and Methods: A descriptive-analytical, cross-sectional survey was conducted between August 2024 and March 2025 at the Imam Khomeini Hospital Complex in Tehran, Iran. A validated questionnaire was administered to a convenience sample of 388 hospital visitors (inpatients, outpatients, and their companions). Data were analyzed using descriptive statistics and multivariate logistic regression.Results: Of 388 participants, 53.6% (208/388) had heard of medical AI, yet only 2.6% claimed full knowledge. A phenomenon regarding conflicting attitudes, which we term the “trust-preference paradox”, was observed: while ~87% expressed conditional trust in AI’s diagnostic and therapeutic capabilities, a significant majority preferred human physicians in disagreement scenarios (Diagnosis: 66.2%; Treatment: 66.8%), primarily due to the desire for personalized care. Preference for AI diagnosis was significantly associated with being male (Odds Ratio [OR] 2.06), having a non-medical/computer science occupation (OR 2.64), and prior AI awareness (OR 1.72). The top perceived advantage was accuracy (46.1%), while the main drawback was lack of empathy (60.1%). A collaborative model where AI assists physicians was favored by 70.6%.Conclusion: Participants demonstrated high trust in AI’s technical potential but strongly preferred human oversight in high-stakes decisions. Successful AI integration in Iran requires “human-in-the-loop” collaborative models and targeted education to bridge the gap between technical trust and clinical preference.Plain Language Summary: Why was the study done? Physicians and hospitals are increasingly using Artificial Intelligence (AI) to help diagnose and treat patients, but public trust is essential for success. We conducted this study to understand the opinions and trust levels regarding medical AI among patients and visitors at a large teaching hospital in Tehran.What did the researchers do and find? We surveyed 388 adults about their knowledge and trust in AI. We found that while a large majority (about 87%) trusted AI’s ability to analyze medical information, most still preferred a human doctor to make the final decision. This was especially true if the AI and the doctor disagreed. Participants emphasized that they value the empathy and personal connection provided by human doctors, which they felt AI lacks.What do these results mean? The results suggest that patients accept AI as a helpful tool but do not want it to replace doctors. They prefer a system where doctors and computers work together, with the doctor remaining in charge. Keeping a human doctor involved in important decisions is crucial to maintaining patient trust.Keywords: machine learning, algorithmic aversion, human oversight, patient trust, physician empathy, human-AI collaboration