Appraisal of AI models in biomedical education: multidimensional testing based on basic definitions, knowledge integration, and curriculum ideology and politics

Artificial intelligence (AI) technology has developed rapidly. AI models have demonstrated the potential to facilitate learning in various technology-aided platforms and application scenarios. The course, “Introduction to Bio-omics”, bringing the cutting-edge content to students, is considered abstract, difficult to understand, and highly dependent on methodology. These features of the course invite more compatible AI models to meet the teaching and learning demands. This study focuses on the learning needs of biomedical students, selecting AI models with different technology platforms and application scenarios. AI models selected are tested in terms of three dimensions (basic definition, knowledge integration and expansion, and curriculum ideology and politics) which are designed for systematic evaluating AI models in biomedical education represented by the “Introduction to Bio-omics” course. The results indicate that there are significant differences in the effectiveness of how AI models assist learning: some models are good at accurately analyzing basic concepts, while others have advantages in complex knowledge associations and clinical thinking training. In addition, the quality of answers to ideological and political questions is greatly affected by model training data. This study provides empirical evidence for biomedical students and educators to choose appropriate AI-assisted tools and explores the optimization directions of AI in biomedical teaching in the future, such as enhancing professional domain adaptability, strengthening medical ethics guidance and so on.