Truace
Subscribe

Only 3% of family medicine clerkships let students use ambient AI, a survey of directors finds

Study title: The Ambient AI Era and Medical Student Education: Insights From Family Medicine Clerkship Directors

A survey of family medicine clerkship directors found that medical students almost never get to use ambient AI, the tools that turn clinician-patient conversations into notes. Of 94 directors who answered the AI questions, only 3 (3.2%) reported that the tool was both available and permitted for students.

Read the study abstract

Background and objective Artificial intelligence (AI) adoption in medicine is rapidly expanding, including ambient AI (AAI), which converts clinician-patient conversations into documentation. Despite growing use among practicing physicians, little is known about student access to AAI or barriers to implementation during medical student clerkships. Methods We surveyed 174 family medicine clerkship directors (CDs) through the Council of Academic Family Medicine Educational Research Alliance network regarding AAI a…

Truvace record2 min read1 sourceEvidence: Emerging · 46/100 — Survey · n = 94 · 1 independent sourcePeer-reviewedTRV-2026-2393 · cite or check it
The Ambient AI Era and Medical Student Education: Insights From Family Medicine Clerkship Directors"Medical laboratories and clinics in Kinwat" by Mouryan, CC BY-SA 4.0.

In brief

Opinion was split. 42.6% of directors said students should not use it at all, 41.5% supported some use and 16% were unsure. Among opponents, 82.5% worried about clinical reasoning and 85% about communication skills.

These are directors' beliefs, not measured effects on students. The authors say the study did not assess whether actual effects on these skills exist, and they urge schools to design curricula that assume the tools will be available, since physicians are adopting them quickly.

Main points

  1. Of 174 surveys sent through the CERA network, 100 directors responded (57.47%); 94 completed the AI questions.
  2. 38.3% said ambient AI was unavailable in their health system, and another 58.5% said it was available but barred for students by the health system, school or clerkship leadership.
  3. Opinions on student use: 42.6% not at all, 33% for select encounters, 2.1% for most encounters, 6.4% for any encounter, 16% unsure.
  4. 53.2% agreed student use would harm clinical reasoning and 55.3% agreed it would harm communication of clinical information; only 28.7% agreed it would improve training.
  5. 74.5% agreed ambient AI will help patient care, and 47.9% agreed students need it because it will become commonplace.
  6. Among directors who opposed student use, 77.5% disagreed that the tool would improve training, versus 15.4% of supporters (P<0.0001).
  7. The authors recommend curricula that assume ambient AI is available and assessing clinical reasoning through encounter transcripts or recordings rather than only written notes.

The problem

In a survey of 94 family medicine clerkship directors, only 3.2% had ambient AI available and permitted for students, and 53.2% to 55.3% agreed student use would harm clinical reasoning or communication skills.

The full story

The questions were added to the 2025 Council of Academic Family Medicine Educational Research Alliance survey of clerkship directors. Respondents averaged 16.5 years since residency, 66% were female and 65 of 94 worked at public schools. Responses were analysed with descriptive statistics and Fisher's exact tests with Benjamini-Hochberg correction.

Cost and policy both play a part. The authors cite $84 to $600 per monthly user for health systems adopting the technology as a likely reason for unavailability, while 42.5% of directors said their health system itself did not permit student use.

Directors were more positive about faculty and patients than about students. Only 15.8% perceived supervising faculty as resistant to ambient AI, and 74.5% agreed it would help patient care, while only 17.0% agreed it would reduce burnout.

The authors warn that students may adopt free or low-cost tools on their own before permission is given, raising privacy, HIPAA and hallucination risks. They recommend enterprise-grade tools, faculty education and pilot programs to test effects on clinical reasoning and communication.

Sources

  1. Peer-reviewedFamily Medicine2026-11-02

The debate