Semantic and Visual Pathways to Artificial intelligence Literacy. Challenges and Lessons Learned in the Medical Domain

Artificial intelligence (AI) enhances medical research through greater accuracy and large-scale data handling. However, many healthcare professionals face methodological and reproducibility challenges due to limited AI training. Visual programming platforms help address this gap by enabling users to create machine learning (ML) workflows without extensive coding. This article analyzes such platforms, identifying key usability and educational challenges. Drawing on our experience developing a visual ML platform that uses an informal semantic representation approach, we propose improvements that leverage semantic representations—such as clearer workflow segmentation, interactive tutorials based on intuitive conceptual definitions, visual aids, and structured methodological checklists. By defining concepts and relationships through semantically inspired visual structures, these enhancements aim to improve user understanding, AI reliability, and research quality in medical contexts. Future research will evaluate their impact on AI education and application.