Dive Brief:
- Physicians’ AI use is moving beyond administrative tasks to more everyday clinical decision support, like searching through medical literature, according to a survey published Thursday by Doximity.
- More than half of physicians said they use AI daily and another third use it at least once a week, the medical professional and technology firm reported. Most physicians also said that AI has taught them clinical knowledge they didn’t already know.
- Still, physicians remained concerned about AI eroding their skills, with more than half saying they specifically worry about how AI use will impact the development of residents, fellows or trainees they supervise.
Dive Insight:
The initial wave of AI adoption was focused on administrative work, especially in supporting clinical documentation — a notorious driver of clinician burnout.
While AI is continuing to help physicians mitigate time-consuming administrative tasks, the Doximity survey of more than 3,300 of the firm’s U.S. doctors shows that physicians are increasingly using it to support clinical decision-making. The share of physicians using AI for clinical decision support nearly doubled to 63%, up from 37% in 2025.
Also, more physicians — 70% from just over half in last year’s survey — are using AI to comb through troves of medical literature.
The survey indicates that physicians may be learning from AI too. A majority (81%) of physicians said that using AI has taught them something they previously didn't know, including updated guidelines, alternative diagnoses or treatment options they hadn't considered. Another 65% said that AI tools have helped with developing their clinical skills and knowledge.
But that support is also raising concerns about the technology's impact on greener providers, with 61% worrying that trainees will rely on AI too much and not develop their own set of clinical skills.
And that fear is not limited to trainees. Nearly half of the physicians surveyed said they were concerned about overreliance on AI, and more than a quarter of physicians acknowledged that skill erosion is the main factor limiting their use of AI.
Physicians are also still worried about AI’s accuracy, with three in four physicians saying they encountered a factual inaccuracy at least once in the past year while using AI for work. More than half also cited inaccuracies as the main reason for reducing their use of AI.
However, these worries may eventually diminish, the survey suggested. Over a quarter of physicians said the effort they put into correcting or double-checking AI output has decreased compared with a year ago.
