# Healthcare Providers’ Perspectives on Generative Artificial Intelligence (GenAI) Adoption, Adaptation, Assimilation, and Use in the United States

## TL;DR

U.S. healthcare providers see generative AI as useful for patient care but face limited training and concerns about job loss and ethics.

## Contribution

The study reveals a gap between clinician interest in GenAI and institutional readiness, emphasizing the need for training and governance.

## Key findings

- 86.9% of clinicians believe GenAI is useful in patient care, with confidence increasing over time.
- Only 42.4% of clinicians have received formal GenAI training, and 23.2% report organizational AI adoption.
- Top benefits include documentation efficiency and error reduction, while barriers include limited AI literacy and fear of job displacement.

## Abstract

What are the main findings?
Most U.S. clinicians in the study perceive GenAI as useful in patient care, with confidence increasing over time, indicating strong momentum for adoption.Formal GenAI training and organizational adoption remain limited, creating a misalignment between clinician interest and system-level preparedness.Clinicians primarily value GenAI for documentation efficiency and error reduction, while barriers center on limited AI literacy and workforce displacement fears.

Most U.S. clinicians in the study perceive GenAI as useful in patient care, with confidence increasing over time, indicating strong momentum for adoption.

Formal GenAI training and organizational adoption remain limited, creating a misalignment between clinician interest and system-level preparedness.

Clinicians primarily value GenAI for documentation efficiency and error reduction, while barriers center on limited AI literacy and workforce displacement fears.

What are the implications of the main findings?
Healthcare organizations must prioritize standardized, role-specific training to convert positive sentiment into safe and effective use.Successful integration will require transparent governance, clinician oversight, and clear accountability to address ethical and trust concerns.Healthcare administrators should intentionally reinvest productivity gains into strengthening patient–provider relationships, improving work–life balance, and reducing clinician burnout.

Healthcare organizations must prioritize standardized, role-specific training to convert positive sentiment into safe and effective use.

Successful integration will require transparent governance, clinician oversight, and clear accountability to address ethical and trust concerns.

Healthcare administrators should intentionally reinvest productivity gains into strengthening patient–provider relationships, improving work–life balance, and reducing clinician burnout.

Background: Generative artificial intelligence (GenAI) is rapidly permeating healthcare; yet, U.S. clinicians still report mixed feelings about its reliability, impact on workflow, and ethical implications. Current data on provider sentiment are needed to guide safe, patient-centered AI implementation in healthcare. Objective: This study aimed to assess U.S. healthcare providers’ perceptions of generative AI adoption, perceived usefulness, training needs, barriers, and strategies for safe integration. Methods: A nationwide, IRB-approved, cross-sectional survey was administered to healthcare professionals using Qualtrics. A convenience sample of clinicians was recruited via professional listservs and e-mail invitations. The 20-page questionnaire captured demographics, GenAI exposure, organizational adoption status, perceived usefulness (5-point scale), barriers, and mitigation strategies. SPSS v27 and Microsoft Excel were used for statistical analysis. Results: Of 130 respondents, 109 completed the core survey (completion rate 83.8%). Participants were 38.5% physicians, 16.5% nurses, 12.8% allied professionals, and 32.2% other providers; 54.2% were women, and 64.8% were ≥50 years. Overall, 86.9% agreed that GenAI is useful in current patient care, rising to 92.9% when asked about future usefulness. Only 42.4% had received formal GenAI training, and just 23.2% reported that their organization had begun adopting AI. The top perceived benefits were improved documentation/clerking (57.0%) and error reduction (49.4%). Dominant barriers included limited AI knowledge (24.7%) and fear of job loss (16.9%). Despite concerns, 72% expressed willingness to support broader GenAI adoption, favoring human oversight (67.1%) and staff training (60.8%) as key safeguards. There were statistically significant findings in perceived AI usefulness by gender (χ2 = 29.2; p < 0.001); organizational adoption of AI (χ2 = 31.6.2; p = 0.047) and where AI is most useful (χ2 = 101.1; p < 0.001) by qualifications; and support for AI adoption by age (χ2 = 18.0; p = 0.02). Conclusions: U.S. clinicians in our survey viewed GenAI as useful but reported limited training and organizational infrastructure needed for confident use while also expressing concerns regarding data privacy and ethical risk. Education programs and transparent, provider-led implementation strategies may accelerate responsible GenAI assimilation while addressing ethical and workforce concerns. Also, health administrators should use the efficiency gains to improve provider–patient relationships and clinicians’ work–life balance while reducing clinician burnout rates.