Majority of doctors worry about AI driving clinical decisions, survey shows
Two out of three physicians are concerned about AI driving diagnosis and treatment decisions, seeing it as a tool better suited for scheduling or other administrative tasks, according to a new Medscape survey.
The big picture: Many clinicians worry current iterations of AI could make flawed recommendations and drive up their legal liability. And like professionals in many other fields, they're also concerned about job security.
Yes, but: Older doctors aged 45 to 64 were more enthusiastic about AI's possibilities, countering the perception that younger clinicians who grew up in the rapid-fire digital age view the technology more positively.
- That's possibly because doctors under 35 have had little time to practice yet and aren't in the best position to assess AI's place in their jobs.
What they found: Cardiologists, anesthesiologists and radiologists were likelier to be enthusiastic about AI's future since they're among the specialties where algorithms showed promise in diagnosis and treatment, Medscape found.
- Family physicians and pediatricians were the most likely to show concern about AI's role in their workplaces.
- 65% of all respondents said they were very or somewhat concerned that AI could be substituted for their clinical judgment and expertise.
- 56% were very enthusiastic or enthusiastic about AI in a support role, like offering diagnosis and treatment options to supplement doctors' recommendations.
Asked about how medical practices already use AI, respondents most often listed researching conditions (13%), electronic health records (10%), administrative tasks (10%) and patient scheduling (9%).
What they're saying: "Physicians want AI to take away the most boring and tedious aspects of their job," Yair Lewis, chief medical officer at AI startup Navina, told Medscape. "This survey found concerns about AI in medicine as a general concept, but then when you start to ask about specific tasks, you see significant enthusiasm."
- He predicted that in another year or two, AI applications could be making diagnoses or predicting a patient's outcomes.
Of note: One in four physicians had negative views about using AI for communicating with patients, echoing earlier concerns about how AI tools could handle private or sensitive information.
- 88% said it was at least somewhat likely that patients who turn to generative AI like ChatGPT for medical information would get at least some misinformation.
- A similar amount are at least somewhat concerned that patients researching their conditions could side with AI-generated information over a doctor's recommendation.
- The survey of 1,043 physicians was conducted from July 12 to Aug. 11 and has a sampling error of 3% at a 95% confidence level.