Preparing tomorrow's physicians for AI-driven healthcare: insights from a study on medical students', interns', and residents' knowledge, attitudes, and educational needs

Artificial intelligence (AI) is increasingly embedded in healthcare delivery, influencing clinical practice, medical education, and research. Despite rapid technological advancement, limited empirical evidence exists on how medical trainees across training levels perceive AI, their readiness to use it, and the educational, ethical, and system-level conditions required for responsible AI integration.

This study aimed to examine medical and dental trainees' knowledge, attitudes, real-world experiences, and educational needs related to AI in healthcare, with particular attention to trust, workflow integration, human oversight, and institutional governance.

An explanatory sequential mixed-methods design was employed. A structured online survey was distributed to undergraduate and postgraduate medical and dental trainees across the United Arab Emirates (UAE) (n = 154). Quantitative data were analyzed descriptively. Semi-structured interviews were conducted with a purposive subsample of participants (n = 16) and analyzed using inductive reflexive thematic analysis following Braun and Clarke's framework. Open-ended survey responses were used to support triangulation. Integration of findings was achieved through a joint display linking quantitative results with qualitative themes and illustrative data.

Survey responses (n = 154) demonstrated moderate awareness of AI applications, limited formal AI training, and strong support for structured AI education. Qualitative interviews (n = 16) revealed four overarching themes: (1) perceived value and normalization of AI, (2) AI as a tool for clinical support and workflow optimization, (3) trust, confidence, and human oversight, and (4) ethical, educational, and system-level preconditions. Trainees viewed AI as an assistive and time-saving tool for learning, research, and selected clinical tasks, while expressing context-dependent trust, emphasizing routine verification, human judgment, and concerns related to accuracy, bias, data security, accountability, and preservation of empathy in patient care.

Trainees recognize the transformative potential of AI but emphasize the need for longitudinal, clinically relevant education, robust ethical governance, institutional support, and continued human oversight. These findings suggest that preparing future physicians for AI-enabled healthcare may require human-centered, ethically grounded, and system-ready approaches that align education, workflow integration, and governance.