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

Area of Science:

• Nursing
• Artificial Intelligence
• Clinical Decision-Making

Background:

• Generative artificial intelligence (GenAI) presents opportunities to enhance nursing clinical decision-making (CDM) and address expertise disparities.
• The Value-based Adoption Model (VAM) provides a framework for evaluating technology adoption, focusing on perceived usefulness, complexity, and helpfulness.

Purpose of the Study:

• To compare nurses' perceptions of GenAI-generated clinical advice against human expert recommendations within a simulated environment.
• To assess the impact of clinical experience on the perceived value of GenAI in nursing CDM.

Main Methods:

• A randomized, double-blind simulation study involving 69 nurses.
• Comparison of ChatGPT-4.0 generated advice (standardized) with expert consensus advice.
• Assessment of perceived usefulness, complexity, and helpfulness using VAM, alongside clinical accuracy.

Main Results:

• No significant differences were found between GenAI and expert support in perceived usefulness (p=.89), complexity (p=.29), and helpfulness (p=.73).
• Both GenAI and expert advice demonstrated similar clinical accuracy in simulated scenarios (p=.45).
• Clinical experience and accuracy negatively predicted Perceived Value (PV), with less experienced nurses showing higher PV for GenAI.

Conclusions:

• GenAI can serve as a comparable support tool to human experts in nursing CDM simulations.
• The perceived value of GenAI is higher among less experienced nurses, aligning with the Novice to Expert framework.
• Future integration requires tailored training and trust calibration to position GenAI as a reliable secondary support tool while preserving clinical autonomy.