When ambient AI documentation tools first rolled into hospital exam rooms a few years ago, the pitch was universal: give overworked doctors their evenings back by letting software listen to the visit and write the note. But a new study out of Emory University’s Rollins School of Public Health suggests the technology has not spread evenly, and the hospitals that need relief the most may be the last to get it.
What the Study Found
Published in the American Journal of Managed Care in February 2026, the research team led by Ilana Graetz, a professor of health policy and management at Emory, analyzed 2,784 U.S. hospitals that run the Epic electronic health record system. They found that 62.6% of those hospitals had adopted some form of ambient AI documentation tool by 2025. Three products dominate the market: Microsoft’s DAX Copilot, Abridge, and ThinkAndor, each used by more than 80% of the hospitals that have adopted any ambient tool at all.
Who Got Left Behind
The disparity is the real story. Graetz’s team found that adoption tracked closely with financial health: hospitals with stronger operating margins, larger patient volumes, nonprofit ownership, and metropolitan locations were significantly more likely to have already deployed ambient scribes. Rural hospitals and systems operating on thin margins lagged well behind. The study also flagged a regional gap, with adoption running lower across the Midwest than in the South.
Why the Money Matters
Ambient AI licensing is not cheap, and hospitals also have to invest in IT integration, legal review of data-sharing agreements, and staff training before a scribe tool can go live safely. Well-resourced systems with dedicated informatics teams can absorb that overhead and move fast; smaller or financially strained hospitals often lack the internal capacity to even evaluate vendors, let alone negotiate favorable contracts. The ambient documentation market itself has been growing quickly, crossing $600 million in revenue in 2025, more than double what it generated the year before, which has only sharpened the competitive advantage of systems that got in early.
The Stakes for Patients and Clinicians
The concern is not merely about convenience. Multiple studies have linked ambient scribes to measurable drops in clinician documentation time and, in some cases, meaningful reductions in burnout. If those effects hold up, hospitals without access to the tools could see their doctors and nurses burn out and leave at higher rates than well-funded peers, further straining care in the communities that already struggle to recruit and retain clinicians. “If these technologies prove effective at reducing burnout and improving care quality, unequal access could widen existing disparities between well-resourced and under-resourced hospitals,” Graetz said, summarizing the core worry driving the research.
A Counterpoint: Not Every Hospital Wants In Yet
Not everyone treats slower adoption as a crisis. Some health IT leaders argue that being a fast follower rather than an early adopter has real advantages, letting smaller systems watch which vendors survive, which contracts include favorable liability terms, and which implementations actually reduce administrative burden versus simply shifting it. Because ambient scribes are currently classified as administrative tools rather than medical devices, they fall outside FDA oversight, and some hospital compliance officers say they would rather wait for clearer governance norms to emerge before committing scarce dollars to a product category still being shaped by early missteps and vendor consolidation.
What Comes Next
Graetz and her co-authors are calling for policy interventions modeled on the support programs that helped narrow the digital divide during the original EHR meaningful-use push over a decade ago, including subsidies, shared implementation resources, and independent cost-effectiveness analysis so smaller hospitals are not simply guessing at return on investment. Industry watchers expect 2026 to be less about proof-of-concept pilots and more about normalization, as ambient AI shifts from novelty to expected infrastructure. The open question is whether that normalization happens for every hospital or mainly for the ones that could already afford to move first. Researchers say tracking adoption by hospital resource level over the next two years will show whether the gap closes or hardens into a permanent two-tier system of documentation support. Graetz’s team has also suggested that vendors themselves could play a role in narrowing the divide, for instance by offering tiered pricing structures scaled to hospital size and payer mix rather than flat enterprise contracts that effectively price out smaller systems from the start. Until that happens, the study’s authors argue, ambient AI risks becoming another example of a genuinely useful clinical technology that reinforces rather than reduces the resource gap between America’s best-funded hospitals and everyone else.
