Nurses’ Attitudes and Experiences With Generative AI in Clinical Practice: Qualitative Systematic Review
Journal of Medical Internet Research ·
Background: The rapid advancement of generative AI (GenAI) holds transformative potential for clinical nursing practice, yet its integration also brings significant risks and uncertainties. Despite growing discourse on AI in health care, frontline nurses’ firsthand experiences, perceptions, and practical needs regarding GenAI remain largely under-synthesized in the existing literature. Objective: This systematic review aimed to synthesize qualitative evidence to clarify the core themes related to clinical nurses’ perceptions, experiences, obstacles, and requirements when using GenAI in real-world clinical settings. Methods: A systematic search was conducted across 10 electronic databases (PubMed, Embase, CINAHL, Web of Science, Cochrane Library, ProQuest Dissertations and Theses, China National Knowledge Infrastructure, VIP Information, SinoMed, and Wanfang) for records published between January 2018 and June 2026. Eligible studies included qualitative and mixed methods studies involving registered nurses that explored their perceptions and experiences of GenAI in clinical practice. Two independent reviewers (MMW and CYZ) performed study selection, quality appraisal, and data extraction, with disagreements resolved by consensus. Quality was assessed using the Joanna Briggs Institute (JBI) Critical Appraisal Checklist for Qualitative Research. Extracted data encompassed study characteristics, participant demographics, and GenAI-related contextual factors. Data synthesis followed Thomas and Harden’s 3-phase thematic synthesis: initial inductive coding generated descriptive themes, which were subsequently integrated with the Technology Acceptance Model for Artificial Intelligence in Nursing (TAM-AIN) framework to develop higher-order analytical themes. Certainty in the evidence was evaluated using the GRADE-CERQual (Grading of Recommendations Assessment, Development and Evaluation-Confidence in the Evidence from Reviews of Qualitative Research). Results: Of the 4350 initially identified records, 13 studies from 6 countries (China: n=6; Saudi Arabia: n=3; United States, Singapore, Australia, and South Korea: n=1 each) met the inclusion criteria, encompassing qualitative data from 234 registered nurses. Three main themes emerged: patterns of nurses’ engagement with GenAI (ranging from passive observation to proactive integration), multidimensional perceived threats to adoption (functional, professional, and accountability-related), and the moderating role of organizational readiness, which influenced whether these threats hindered or facilitated GenAI uptake in practice. The GRADE-CERQual assessment showed moderate to high confidence in all findings. Conclusions: Nurses’ engagement with GenAI ranges from passive observation to proactive integration, accompanied by legitimate concerns about functional reliability, professional role displacement, and accountability ambiguity. The moderating role of organizational readiness further highlights that training, tool optimization, and institutional governance remain critical yet often unmet. These findings necessitate coordinated efforts across workforce development, technological design, and regulatory frameworks. Future research should prioritize the development of clinically validated, context-aware GenAI systems that incorporate human-in-the-loop oversight, ensuring that efficiency gains do not compromise the delivery of compassionate, person-centered care. Trial Registration: PROSPERO CRD420251250664; https://www.crd.york.ac.uk/PROSPERO/view/CRD420251250664