Building the Blueprint for Medical Educators of West Bengal: A Delphi Study on developing consensus-based competency framework for Large Language Models

Background: The increasing use of Large Language Models (LLMs) demands an official change in medical education preparation. The Indian context lacks a consensus model to inform the training of medical educators in LLM competencies. This deficit risks unprepared faculty and graduates within an AI-augmented healthcare system. The objective of this study was to develop a consensus-based list of core competencies for medical educators. Material and Methods: A descriptive, observational study using a modified Delphi technique was conducted over six months, April 2025 to October 2025. A twenty-five-member expert panel, purposively selected and stratified by medical education discipline, was used. All participants had considerable experience and prior AI training. Three rounds of online questionnaires were used, with the consensus for an item’s inclusion was decided by a median score of 4.0 or more and an Interquartile Range (IQR) of 1.0 or less. Results: Core competency items showed exceptionally high levels of consensus (31/31). Thirty-one core competency items met prominent levels of consensus (IQR = 1.0 or 0.0). The competencies of greatest priority revolved around three pillars: Ethical and Safety Imperatives, Practical utility in academic roles, and a supportive Institutional Framework. Conclusion: This research forms a solid, consensus-built model of LLM literacy. The results provide further clear guidance for curriculum developers to shift medical education away from technical theory and toward practical skills. Ongoing curricular change and formal faculty development are required.

Keywords: Large Language Models, Medical Education, Curriculum Development, Delphi Technique, Generative AI.