Bridging Expertise with Algorithms: Evaluation of Generative AI in Nursing Decisions

Integrating generative artificial intelligence (GenAI) into nursing clinical decision-making (CDM) offers potential to bridge expertise gaps. This randomized, double-blind simulation (n=69) compared nurses’ perceptions of GenAI-generated clinical advice versus human expert recommendations. Expert advice was derived from a structured consensus process, while GenAI output (ChatGPT-4.0; Temperature 0.2) was visually and structurally standardized for comparability. Guided by the Value-based Adoption Model (VAM), we assessed perceived usefulness, complexity, and helpfulness (defined as willingness to modify decisions). Results showed no significant differences between GenAI and expert support in usefulness (p=.89), complexity (p=.29), and helpfulness (p=.73). Both sources demonstrated similar clinical accuracy within the simulated scenarios (p=.45). Multiple regression revealed that clinical experience (B = - 0.042, p<.05) and clinical accuracy (B = -1.694 p<.05) were significant negative predictors of Perceived Value (PV). These findings are consistent with Benner’s "Novice to Expert" framework, suggesting that GenAI PV is most recognized by less experienced nurses who rely more on external support. Future implementation should focus on tailored trust calibration and professional training to integrate GenAI as a reliable secondary support tool without compromising clinical autonomy.