Doctors ask AI how to treat patients more than what they have

The diagnosis is the scene everyone remembers. Then the patient asks what happens now, and the harder part begins: which treatment, and whether it clashes with the pills already in the bag.

That second question is the one clinicians bring to AI more often. A JAMA study published online October 15, 2026 looked at OpenEvidence, an AI clinical information platform, where "952 008 clinicians (58.5% female), including 556 200 physicians, submitted 271 million queries during the 2.5-year study period." OpenEvidence supplied the data, and three of its employees co-authored the study, IBM reports. A language model sorted a sample of 38.8 million queries by purpose.

In the abstract's words: "Most queries addressed clinical decision-making (51.9%), particularly treatment and management (33.1%), whereas diagnosis accounted for 13.9% of queries." The Key Points give those two as shares "of physician queries." Treatment outnumbers diagnosis. Neither is a majority.

Doctors also ask outside their own field. "Queries addressing topics outside physicians' own specialties ranged from 35% among oncologists and hematologists to 88% among emergency department physicians." For an emergency physician, outside the specialty is more or less the job description.

The question doctors bring to the machine is less often what is this and more often what now. Whether the machine's answers were any good is a different study, and this is not that study.