Don't let the bot play doctor! AI gets early diagnoses wrong 80% of the time

A new study published in JAMA Network Open shows that leading AI models fail early differential diagnosis in over 80 percent of cases, even though they perform well when given full medical information for final diagnoses. Researchers tested 21 off‑the‑shelf models across 29 clinical vignettes and found that all struggled most at the stage where uncertainty is highest and clinical reasoning matters most. Experts warn that LLMs project confidence without reliable reasoning, which can mislead patients and delay proper care. The findings reinforce that AI tools are not ready for frontline diagnostic decision‑making.

Read the full story on The Register