AI Adoption in US Hospitals: Workload, Finances, and Location Play Key Roles

As of June 2025, approximately 62.6% of U.S. Hospitals utilizing the Epic electronic health record (EHR) system had adopted ambient AI applications, according to a study published in the American Journal of Managed Care. Adoption rates vary significantly based on hospital characteristics, including workload, operating margins, location, and ownership.

Growing Demand for AI in Healthcare

Researchers at Emory University, led by Ilana Graetz, PhD, observed a rapid increase in the use of ambient AI documentation tools in clinical settings. The study aimed to systematically understand which hospitals were implementing these technologies and the reasons behind their decisions. “Given the potential for these tools to reduce clinician burnout and improve care quality, we wanted to leverage available data within the Epic system…to better understand the landscape of adoption,” explained Dr. Graetz.

Key Factors Influencing AI Adoption

Workload

Hospitals with higher workloads demonstrated a significantly greater probability of adopting ambient AI. Specifically, hospitals in the highest workload quartile (quartile 4) had a 73.1% predicted probability of adoption, compared to 61.7% in the lowest quartile (quartile 1).

Operating Margins

Financial health similarly played a crucial role. Hospitals with the highest operating margins (quartile 4) showed a 67.6% predicted adoption rate, while those with the lowest margins (quartile 1) had a rate of 58%.

Hospital Size

Larger hospitals were more likely to adopt ambient AI compared to medium-sized hospitals, with adoption rates of 60.9% versus 52.2%, respectively.

Location

Metropolitan hospitals exhibited higher adoption rates (64.7%) than nonmetropolitan hospitals (54.3%).

Ownership

Nonprofit hospitals had significantly higher adoption rates (70.2%) compared to government hospitals (45%). For-profit hospitals demonstrated the lowest adoption rates at 28.8%.

Regional Differences

Hospitals in the Midwest had a lower adoption rate (54.9%) compared to those in the South (69.5%). The Northeast and West regions showed adoption probabilities similar to the South.

Popular AI Tools

The study identified DAX Copilot, Abridge, and ThinkAndor as the most frequently used ambient AI tools among Epic hospitals, collectively capturing over 80% of the market share. Researchers anticipate this landscape will evolve as Epic introduces its own integrated AI charting solution.

Implications and Future Research

Dr. Graetz emphasized the importance of understanding access gaps. “Clinicians working in well-resourced, nonprofit systems, particularly in metropolitan areas, are more likely to have access, whereas clinicians in for-profit hospitals, rural settings or financially constrained systems may face longer delays, potentially widening disparities in documentation burden, and burnout.”

Future research should focus on evaluating actual clinician usage, long-term engagement, and the impact of ambient AI on patient outcomes, including quality, safety, clinician burnout, and job satisfaction.