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

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: To assess U.S. healthcare providers’ perceptions of GenAI 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<.001); organizational adoption of AI (χ²= 31.6.2; P=.047) and where AI is most useful (χ²=101.1; P<.001) by qualifications; and support for AI adoption by age (χ²= 18.0; P=.02). Conclusions: U.S. clinicians in our survey viewed GenAI positively but lacked the training and organizational infrastructure needed for confident use. Structured 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.