Background: Whether AI assistance relates to care quality is uncertain.
Objective: To compare initial AI recommendations against physicians' final recommendations when physicians had access to the AI's suggestions.
Design & Setting: A retrospective cohort study at Cedars-Sinai Connect, an AI-assisted virtual urgent care clinic with intake questions via structured chat.
Patients: 461 physician-managed visits involving adults with respiratory, urinary, vaginal, eye, or dental symptoms, from 12 June to 14 July 2024.
Results: AI and physician recommendations matched in 262 visits (56.8%). AI recommendations were rated optimal more often (77.1%) than physician decisions (67.1%). Quality was equal in 67.9% of cases, better for AI in 20.8%, and better for physicians in 11.3%.
Conclusion: When recommendations differed, AI was more frequently judged higher quality — stronger at identifying critical red flags and supporting guideline-adherent care — while physicians better adapted to changing information during consultations. AI may have a role in assisting physician decision making in virtual urgent care.