Primary care physicians underestimate how comfortable patients are with AI making clinical decisions that a provider can change or override, according to a Nov. 9 report from the West Health-Gallup Center on Healthcare in America.
When AI makes diagnosis and treatment decisions that a provider can override, 56% of patients said they would be comfortable. Only 36% of physicians believe most of their patients would feel the same, a 20-percentage-point gap.
The gaps were smaller for imaging review (53% of patients vs. 42% of physicians) and medication decisions (49% vs. 42%).
Patient comfort depends on provider oversight. Without provider review, comfort drops to 9% for diagnosis and treatment decisions, 13% for reviewing diagnostic images and 11% for medication decisions.
The form of oversight also matters. For diagnosis and treatment, patients were more comfortable with AI making decisions a provider can override (56%) than with AI assisting a provider who makes the decision (38%).
When AI only assists, physician estimates came within two points of patient responses for diagnosis and treatment and for medication decisions. Physicians still underestimated patient comfort with AI imaging review by 11 points.
Physicians are also adopting AI faster than they are being trained on it. Eighty-nine percent reported using AI for work in the past year, but only 20% had received any formal training in using AI in healthcare settings. That share was 30% among physicians in hospital or health system practices, 16% in physician groups and 11% in independent practices.
The findings come from an Aug. 12-24 survey of 5,330 U.S. adults who had seen their primary care provider in the past year and a separate opt-in survey of 1,007 primary care physicians, also conducted in August. It is the first time the center has paired patient and provider surveys on AI.
The researchers said misjudging patient comfort could cause healthcare organizations to miss beneficial AI applications that align with patient preferences.
