2026/04/29 by Crystal Uminski · 1 voice
Biochemistry, Genetics and Molecular Biology · Social Sciences · Psychology · #Genetics, Bioinformatics, and Biomedical Research #Science Education and Pedagogy #Visual and Cognitive Learning Processes
paper · doi:10.1128/jmbe.00321-25
openalex publication_date 2026/04/29 · openalex created_date 2026/04/30 · openalex updated_date 2026/05/21
Artificial intelligence (AI) has rapidly improved in its ability to produce polished and realistic visual representations of biological structures and processes, yet most AI-generated images still contain factual inaccuracies, inconsistencies, or errors. These types of errors that often make AI images "bad" representations of biology can make AI images powerful pedagogical tools for teaching visual literacy. As biology education relies on diagrams to model structures, processes, and systems that cannot be directly observed, students need to develop skills interpreting, evaluating, and critiquing the visual representations of the unobservable phenomena they encounter in their biology courses. The error-prone and uncanny qualities of AI-generated images can provide rich opportunities for active learning in which students identify inaccuracies, articulate why AI images are incorrect, and refine their understanding of the visual conventions in biology diagrams. The flaws and inaccuracies that make AI-generated biology diagrams "bad" can make them quite good for fostering critical thinking and deep conceptual understanding in biology learners. This perspective outlines practical strategies for incorporating AI images into active learning and assessments in biology courses and offers guidance for generating AI images to use as teaching tools.