LO
CapabilityCognitive automation2026-08-11
Google Translate's voice mode had linguistic or clinical errors in 33.3% of real clinic speech segments against 4.8% for qualified medical interpreters, a study found
Interpreteroccupation page →
Event date / reported
2026-08-11
Evidence stage
CapabilityA demo, benchmark or paper shows the task can be done. Updates what the technology can do — not what employers will do.
Tasks this bears on
Medical interpreting
Interpreting between patients and clinicians, in person or remotely.
Still human-led≈ Platform inference
Where this applies
United States, a clinic setting, six languages. Using real clinical speech segments, the authors report that Google Translate in speech/voice mode had significantly higher linguistic and clinical error rates than qualified medical interpreters (33.3% vs. 4.8%), and that errors were worse for less commonly spoken languages (42.9% vs. 14.3%). It is a small exploratory study of one app.
What this means
In real clinic speech, a widely used app made errors in about one segment in three, against one in twenty for qualified interpreters — the gap that keeps interpreters in clinics.
What it does not yet show
A small study of one app; newer tools may do better.
What you can check
Open the Healthcare article (PMC13512232) and find "33.3% vs. 4.8%".
Does it change the assessment?
No — and this stage does not move it either. A "Capability" record is real evidence, but it does not upgrade a task judgement on its own. The 1 linked judgement above stand where they were.
Source
Healthcare (MDPI) 14(16):2497 — comparison of Google Translate speech/voice mode with qualified medical interpreters, Georgia State University authors (published 11 August 2026) · verified 2026-09-30 · Claude (VOLO agent) · interpreted 2026-09-30 · Claude (VOLO agent)
Primary source — published by the party that did this, or the authority of record. No co-signature needed.