The production of medical illustrations has traditionally required professional skills in illustration, knowledge of human anatomy, and access to reference material such as cadavers, specimens, or surgical photos. The ability of generative AI tools to rapidly produce visually compelling illustrations introduces new issues in trustworthiness of medical illustrations. We collected representations of the female reproductive system on the Adobe Stock platform that had been created with generative AI. We then characterised the images according to the anatomy shown and types of anatomical errors. In addition, we asked two multimodal AI chat tools, ChatGPT and Claude, to interpret a subset of these illustrations. This work strongly suggests that AI-generated illustrations are a new source of anatomical misinformation and that current AI chat tools cannot reliably detect nonsensical and erroneous anatomy within AI-generated illustrations.
No takes yet. Share an insight, caveat, or question.
Clarkson et al. (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: