Introduction to Ambient AI in Healthcare
Electronic health records (EHRs) have revolutionized healthcare documentation across the United States, with widespread adoption among hospitals nationwide. However, this digital transformation has introduced an unexpected challenge: documentation burden. Clinicians now spend nearly two hours documenting patient encounters for every hour of direct clinical time, creating significant workflow disruptions and contributing to physician burnout.
Ambient artificial intelligence (AI) technologies have emerged as a promising solution to this growing problem. These innovative tools capture clinician-patient conversations in real-time and automatically generate draft clinical notes for physician review, potentially reducing administrative workload while improving patient-provider communication. Despite their promise, little is known about how rapidly these technologies are spreading across US healthcare systems or which hospitals are adopting them.
The Documentation Burden Challenge
Impact on Physician Burnout
Documentation burden represents more than just an inconvenience—it contributes significantly to physician burnout, which decreases productivity and increases the likelihood of medical errors. Healthcare professionals have explored multiple approaches to address this challenge, including medical scribes, dictation tools, and workflow optimization strategies. However, traditional solutions often fall short of providing comprehensive relief.
Emergence of AI-Powered Solutions
Ambient AI documentation tools represent the next generation of solutions designed to alleviate clinical documentation burden. Preliminary research suggests these technologies may reduce documentation time, improve after-hours work, and enhance perceived burden among clinicians. Early evidence also indicates potential benefits for patient-provider communication, suggesting these tools could play a crucial role in mitigating physician burnout while maintaining high-quality clinical documentation.
Study Methodology and Data Sources
Research Design and Hospital Identification
This cross-sectional study examined ambient AI adoption among US hospitals using Epic EHR systems as of June 2025. Researchers analyzed 6,561 unique hospitals from Medicare enrollment data, identifying 2,784 hospitals (42.4%) as Epic inpatient EHR users through a combination of American Hospital Association (AHA) Annual Survey data and domain-level matching approaches.
Ambient AI Tool Classification
The study utilized Epic Showroom, Epic’s marketplace for third-party applications, to identify 12 ambient AI products under the “Ambient Voice Recognition” category. The three most commonly adopted tools—DAX Copilot, Abridge, and ThinkAndor—together account for more than 80% of all ambient AI implementations among Epic hospitals.
Predictors and Variables Analyzed
Researchers examined multiple factors potentially influencing adoption patterns, including staffing-adjusted workload, operating margins, hospital size, teaching status, metropolitan location, ownership type, Census region, case mix index, Disproportionate Share Hospital (DSH) status, and Critical Access Hospital designation.
Key Findings on AI Adoption
Overall Adoption Rates
Among the 2,784 US hospitals using Epic EHR systems, 1,744 hospitals (62.6%) had adopted at least one ambient AI tool by June 2025, demonstrating rapid diffusion of AI-enabled documentation support across the healthcare landscape. This widespread adoption suggests that healthcare organizations recognize the potential value of these technologies in addressing documentation challenges.
Workload and Volume Patterns
Hospitals that adopted ambient AI demonstrated significantly higher staffing-adjusted workload compared to non-adopters, with median total volume per full-time equivalent employee of 208.0 versus 184.0 (P < .001). Adopting hospitals also processed substantially higher outpatient volumes, with median annual outpatient visits of 131,405 compared to 82,221 among non-adopters (P < .001).
Financial and Organizational Factors
Operating Margin Influence
Financial performance emerged as a critical factor in ambient AI adoption. Hospitals with stronger operating margins showed higher adoption rates, with adjusted probabilities reaching 68.7% in the third quartile and 67.6% in the fourth quartile, compared to just 58.0% in the first quartile. These findings suggest that financial capacity plays a crucial role in enabling hospitals to invest in innovative AI technologies.
Hospital Size and Metropolitan Status
Larger hospitals demonstrated higher adoption rates, with 12.9% of adopters classified as large facilities (400+ staffed beds) compared to 7.9% of non-adopters. Metropolitan hospitals also showed significantly higher adjusted adoption probability (64.7%) compared to non-metropolitan hospitals (54.3%, P = .012), highlighting potential urban-rural disparities in access to advanced healthcare technologies.
Geographic and Ownership Disparities
Ownership Type Variations
Nonprofit hospitals exhibited substantially higher ambient AI adoption rates compared to other ownership types. After multivariable adjustment, nonprofit hospitals showed a 70.2% adoption probability, significantly higher than government hospitals (45.0%, P < .001). Surprisingly, for-profit hospitals demonstrated lower adoption probabilities (28.8%), potentially reflecting different organizational priorities and tolerance for implementation costs when financial returns remain uncertain.
Regional Adoption Patterns
Geographic variations persisted even after accounting for other hospital characteristics. Hospitals in the Midwest showed lower adoption probabilities (54.9%) compared to Southern hospitals (69.5%, P = .005), suggesting that market factors, vendor penetration, and health system concentration may contribute to regional variation in technology adoption.
Policy Implications and Future Directions
Addressing Adoption Disparities
The observed adoption patterns raise important questions about equitable access to beneficial healthcare technologies. If ambient AI genuinely improves clinician efficiency and care quality, uneven adoption could contribute to widening performance and outcome disparities across hospitals. Policy makers may need to consider targeted approaches similar to previous federal EHR adoption incentives to ensure broader access to effective AI technologies.
Implementation Support Strategies
Cost-effectiveness evidence and implementation supports—such as shared services, scalable pricing models, and technical assistance—could help reduce barriers for financially constrained hospitals. Analogous to Medicare and Medicaid EHR incentive programs, strategic policy interventions might accelerate adoption while monitoring for unintended increases in healthcare disparities.
Limitations and Conclusions
Study Constraints
This analysis focused exclusively on Epic EHR hospitals, potentially limiting generalizability to organizations using other vendors with different integration pathways and pricing models. Additionally, the observational design means results should be interpreted as associations rather than causal effects. Hospital characteristic data drawn from 2023 AHA responses may not fully reflect 2025 conditions.
Future Research Directions
Additional studies are urgently needed to evaluate downstream effects on care processes and patient outcomes, as well as to identify potential unintended consequences related to documentation quality, equity, and safety. Future work should identify modifiable barriers and facilitators to ambient AI adoption and assess long-term impacts on care quality and patient outcomes across diverse hospital settings.
Conclusions
Ambient AI adoption has achieved widespread penetration among US hospitals using Epic EHR systems, with nearly two-thirds implementing these technologies by 2025. Adoption patterns reveal significant disparities based on financial performance, ownership type, geographic location, and organizational characteristics. Strategic policy interventions may be necessary to ensure that AI technology adoption does not inadvertently widen existing care quality gaps, particularly affecting smaller, resource-constrained hospitals serving vulnerable populations.
