AI Meets Assessment: Evaluating the Efficacy of Language Models in Medical Question Generation

This work explores the intersection of artificial intelligence and educational assessment, focusing on how advanced language models can be harnessed to automatically generate high-quality medical questions. With the growing demand for scalable and adaptive assessment tools in medical education, this session evaluates the efficacy, accuracy, and relevance of AI-generated questions compared to human-authored ones. Key topics include the use of transformer-based models (e.g., T5, GPT) for question generation, the challenges in maintaining clinical validity and pedagogical value, and the methodologies used to evaluate question quality in terms of difficulty, coverage, and cognitive depth.