Doctors report feeling better with AI scribes. The measured time savings are smaller than many of them perceive, and the work of checking the AI is still unpriced.
In a 2026 study of 252 physicians in the American Journal of Managed Care, 86.5 percent said an AI scribe reduced their documentation time. The authors also measured documentation time directly. They found no significant association between the change a physician perceived and the change the record showed (odds ratio 0.975, 95 percent confidence interval 0.94 to 1.01).
That does not mean the scribes did nothing. In the same study, measured documentation time fell from 22.2 to 17.3 minutes per scheduled hour. The measured saving was real, though the study cannot isolate the scribe as the cause. It is just not the same thing as the saving physicians felt.
I have spent decades in GI practice and now work as a hospitalist. I want to be fair to the technology, so I will start with what it does well.
Real physician voices, twice a week
Free, and one click to unsubscribe.
The doctors feel better
An open-label randomized crossover trial of two scribes, reported in the Journal of the American Medical Informatics Association in 2026, found that both lowered burnout scores. One caution: The trial had no arm without a scribe, and 136 of the 160 enrolled physicians were analyzed. UW Health and Kaiser Permanente have reported similar benefits, the first in a news release and the second in an observational, self-reported survey.
I count this as a real benefit. A physician who finishes the day less drained is a better physician, and a health system that helps with that has done something worth doing.
The measured picture
The time savings are modest. A large multicenter analysis in JAMA, as reported by MDLinx in June 2026, found about 16 fewer minutes of documentation per eight-hour shift among 1,809 clinicians. I have not seen the primary paper, so treat that figure as reported.
After-hours results are mixed. A Penn Medicine pilot of 46 clinicians reported 30 percent less after-hours time in a news release. The large JAMA analysis, as reported, found no significant drop in after-hours work.
Inbox drafting looks less promising. A University of California San Diego randomized study found that AI-drafted replies did not reduce response time. A five-week Stanford quality improvement study of 162 clinicians, with no comparison group, found no change in the time to act, write, or read.
Where the time goes
Medical Republic reported in April 2026 that time saved on documentation may be reallocated to other work, and warned that health systems may convert it into more visits. If so, the minutes do not return to the physician. They return to the schedule.
That matters because the scribe’s promise was time. If the clock barely moves and the saved minutes go back into the schedule, what remains is how the day feels. That is worth having, but it is not what most purchase orders were written to buy.
The strongest objection
The strongest objection is that I am measuring the wrong thing. If burnout falls, minutes may not be the right yardstick, and a scribe that makes the day feel lighter has earned its place even if the clock barely moves.
I concede most of that. My argument is narrower. Whatever the scribe does for well-being, the physician still reviews what it wrote, and that review is work. A 2026 review of 21 studies in the Journal of Medical Internet Research found it unclear whether scribes reduce cognitive burden or shift it to verification, though it covered many kinds of AI.
The unpriced part
I found no study that totals review time across the AI tools a physician now uses, or that puts a dollar value on it. Seni Kamara, writing in HealthManagement in May 2026, made a similar point: Direct measurement of the verification burden has not been published.
The next claim is my inference, not a finding. If the review step is real and no payer prices it separately, hospitals may be absorbing a cost I found no one measuring or paying for. Chi-way Li is an analyst who tracks medical AI payment. Li’s verification tax, which he calls a working explanation, is the idea I am testing.
An ask
I am asking one thing of readers who use an AI scribe or inbox drafting tool. For one week, count the minutes you spend reading, correcting, and signing what the AI produced. Write the number down.
If your hospital bought the tool on a promise of saved time, that number is the part of the bargain I found no one measuring. I lay out the evidence on scribes and inbox drafts, what each study did and did not show, and the gap in the literature in “The Unpriced Check” on The Renaissance Physician Substack.
Brian Hudes is a board-certified gastroenterologist and hepatologist with more than thirty years of clinical experience and a recipient of his specialty board’s thirty-year certification award. He built and ran a private gastroenterology practice for twenty years, then spent more than a decade in hospital-based gastroenterology, most recently as chief of gastroenterology and medical director of GI and endoscopy at a 550-bed Level I trauma center in Pensacola, Florida. He is an assistant professor of medicine at Florida State University College of Medicine.
Those two halves of a career shape the way he writes about medicine and clinical technology. Outpatient practice and hospital-based care fail their clinicians differently, and many systems are designed without a meaningful understanding of both environments.
Dr. Hudes has been building clinical software alongside his medical practice for decades. In 1995, during his gastroenterology fellowship, he co-developed one of the first Windows-based endoscopy reporting systems in the United States. He is now founder and chief executive officer of AiMOS Systems Corporation, where he is developing an AI-native clinical platform for outpatient and hospital-based care. AiMOS has been accepted into NVIDIA’s Inception program.
He writes about the architecture of clinical information, administrative cost growth, physician workforce shortages, board certification policy, and the widening gap between what clinicians need and what the health care technology industry builds. He also publishes The Renaissance Physician, an independent publication examining medicine, technology, physician workforce, and the structure of modern clinical practice.
