Redefining Mentorship in Medical Education with Artificial Intelligence: A Delphi Study on the Feasibility and Implications.

Abstract

In the dynamically evolving field of medicine, mentorship is crucial for educating students, and Artificial Intelligence (AI) potentially revolutionizes this process through automated and data-enhanced guidance. This study aims to investigate AI's potential in mentoring medical students by collecting expert opinions, assessing its potential benefits and limitations, and developing a consensus-driven framework for the effective integration of AI-based mentorship into medical education. Specifically, it addresses ethical concerns such as data security, algorithmic bias, and the potential for reduced human interaction. Using a structured online Delphi technique, this interdisciplinary research involved 27 experts in medical education and AI to investigate the intersection of AI with medical mentorship. The study employed both qualitative (e.g., expert interviews) and quantitative (e.g., survey data) research methods, with consensus measured via descriptive and inferential statistics, including Fleiss' kappa and the Intraclass Correlation Coefficient (ICC). Detailed methodological steps, including the selection criteria for experts and the iterative feedback process across the four Delphi rounds, were meticulously followed to ensure robust consensus building. Conducted over four rounds, the Delphi technique achieved substantial consensus among panelists regarding the AI mentors' capabilities and the critical aspects requiring attention, with a kappa value of .79 ([.73-.85]) and high reliability (ICC=.873). The study also compared traditional mentorship roles with those enhanced by AI, highlighting areas where AI can complement and extend human mentorship rather than replace it. Panelists recognized AI mentors' potential to enhance learning processes, while also identifying limitations in areas requiring deep human judgment, emphasizing the need for careful application. AI mentors can significantly guide students across various aspects of medical training, from career planning to achieving academic goals, through personalized learning experiences. They hold promise for improving clinical skills and decision-making abilities through real-time feedback and adaptive learning modules. However, their limitations and the potential risks of overreliance necessitate balanced and cautious application. Ethical considerations, such as ensuring data integrity and preventing bias, are paramount in the deployment of AI mentors. These insights advocate the strategic implementation of AI mentors in medical education, suggesting phased integration and interdisciplinary oversight to harness their full educational potential while mitigating possible drawbacks. Furthermore, the study proposes a hybrid mentorship model that combines AI-driven insights with human empathy and ethical oversight to create a more comprehensive and effective mentorship framework. This study lays the groundwork for future research into the optimal integration of AI in medical mentorship, ensuring ethical standards and maximizing educational benefits, thereby fostering a more effective and humane educational environment.