After a quiet start to July the literature suddenly delivered a textbook contrast:
a randomised clinical trial from China demonstrating that GPT-simulated patients can lift history-taking scores by thirteen points;
and an anatomy-education study revealing that popular text-to-image generators still mis-count ribs and blur sulci.
Together they map the road ahead for educators hoping to harness large-language-model power without losing anatomical fidelity. Subscribe now for links to these studies and to read my deep-dive into the original publications.
