A time-motion study just published in JMIR Medical Informatics followed nine clinicians at a Singapore academic medical center, with and without an ambient AI scribe. Documentation time per visit dropped 15 percent. Eye contact with patients rose almost 11 percent. Total visit duration did not change.
That last number is the one vendors rarely put on a slide. The pitch behind most ambient scribe deals is capacity: less charting, more patients seen. STAT reported in April that a larger multi site study found the same pattern, modest time savings and inconsistent use, not the throughput jump the business case assumes.
What actually moved is where clinicians spend the minutes they get back. They look at the patient more. They do not see the next patient sooner. For a system built on volume, that is not a failure of the technology. It is a mismatch between what the tool changes and what the contract pays for.