Do diverse patients trust AI scribes? Not equally.
A large, representative survey of 2,675 U.S. adults found that when given a choice between an AI clinic and a human specialist, the population was almost evenly split — 47.1% chose AI, 52.9% chose a human doctor [1]. But this average hides stark differences: Black respondents were 27% less likely to choose AI than White respondents (odds ratio 0.73), while Native American respondents were 37% more likely to choose it (odds ratio 1.37) [1]. Older adults and politically conservative or religious individuals were also significantly less likely to opt for AI [1]. This means that even if AI scribes work technically, patient buy-in is not uniform — and some groups may resist them, potentially affecting care engagement.
AI scribes struggle with non-English speakers and cultural nuance.
In a qualitative study of 22 physicians using ambient AI scribes across primary care and specialties, the most frequently cited barrier was 'limited functionality with non-English speaking patients' [3]. Physicians reported that the AI tools often failed to accurately capture or interpret conversations in languages other than English, which is a critical gap for clinics serving immigrant or multilingual communities. This finding is echoed by the survey data showing that reassurance about the AI clinic avoiding racial or financial biases did not significantly increase uptake overall [1], suggesting that simply promising fairness isn't enough — the technology itself must perform equitably across languages and cultures.
Does AI scribe documentation capture diverse health conditions equally?
A pilot study of 49 oncology physicians using an AI scribe (Deep Scribe) over 4,500 patient visits found that the AI actually captured more non-cancer diagnosis codes than manual coding — an average of 4.1 codes per patient versus 3.0 [4]. However, the same study showed that cancer diagnosis codes were better captured by manual coding [4], meaning the AI may systematically under-detect certain conditions. This raises a concern: if AI scribes are trained on datasets that underrepresent certain diseases or patient demographics, they could introduce documentation bias that affects billing, treatment plans, and health equity. The randomized trial of 238 physicians across 14 specialties also noted 'occasional inaccuracies' in both AI scribe platforms tested (DAX and Nabla), with clinicians rating accuracy around 2.7–2.8 on a 5-point scale [2] — acceptable but not flawless.
About These Sources
This answer is built on 5 peer-reviewed studies — published from 2023 to 2025, 4 from 2024 or later, 3 in Q1 journals, collectively cited 185 times — selected as the most relevant from 5 studies that passed quality screening, drawn from 28 papers retrieved from a database of over 500 million.
Sources used in this answer
Diverse patients’ attitudes towards Artificial Intelligence (AI) in diagnosis
In a large, representative survey (n=2,675), Black respondents were 27% less likely to choose AI over a human doctor than White respondents, while Native Americans were 37% more likely; older, conservative, and religious individuals also showed lower AI uptake.
Ambient AI Scribes in Clinical Practice: A Randomized Trial
In a randomized trial of 238 physicians across 14 specialties, both AI scribes (DAX and Nabla) reduced burnout and task load, but clinicians reported 'occasional inaccuracies' (rated ~2.7–2.8 on a 5-point scale), and only Nabla significantly reduced documentation time.
Physician Perspectives on Ambient AI Scribes
In a qualitative study of 22 physicians, the most common barrier to AI scribe adoption was limited functionality with non-English speaking patients, though most physicians felt the tool improved patient engagement and reduced workload.
Use of ambient AI scribing: Impact on physician administrative burden and patient care.
In a pilot with 49 oncology physicians and over 4,500 AI-documented visits, the AI scribe captured more total diagnosis codes than manual coding (4.1 vs 3.0 per visit) but under-captured cancer-specific codes, suggesting potential documentation bias.
Impact of using an AI scribe on clinical documentation and clinician-patient interactions in allied health private practice: perspectives of clinicians and patients.
In a mixed-methods study of 119 allied health professionals and 157 patients in Australian private practice, AI scribes reduced documentation time and burden, and patients reported comfort with the technology, though some wanted more information on data security.