“… 6 , 15 , 16 , 17 Studies that have evaluated AI tools in clinical scenarios often fail to capture the nuances of real-world decision-making, where single best solutions are rare. 18 , 19 Rather, clinical judgments frequently hinge on the principle of what a competent peer would consider a reasonable course of action. 20 To our knowledge, AI tools have not been sufficiently evaluated in such contextual conditions.…”