Grounded in social cognitive theory, this study aims to develop a research framework centered on artificial intelligence (AI) self-efficacy to investigate the factors influencing health-care workers’ usage intention of AI systems.
This research used an online questionnaire with 210 valid questionnaires collected from hospital workers serving a regional teaching hospital in central Taiwan. A partial least squares-structural equation modeling (PLS-SEM) and artificial neural network (ANN) approach were used to analyze the users’ responses. First, PLS-SEM was used to validate the model and hypotheses; in the following stage, the ANN approach was used to rank the importance of each influencing factor on the usage intention of AI systems.
This study identified the associations among environment (others’ encouragement, usage and support), person (users’ AI self-efficacy, personal outcome expectations, positive affect and users’ AI anxiety) and behavior (usage intention). Moreover, the other model using ANN validated the results of the SEM analyses that affect, outcome expectations and AI anxiety were essential influences on AI system usage intention.
In the health-care field, AI anxiety is mainly manifested in a multifaceted unease, including doctors’ and patients’ concerns about the accuracy and safety of AI in diagnosis and surgery, health-care workers’ fears that AI may replace their work or weaken their professional authority and technological dependency and privacy issues that deepen trust and security concerns. Therefore, when introducing AI systems into administrative processes, hospital administrators can develop measures that prioritize improving hospital workers’ affect and outcome expectations for using AI systems and alleviating workers’ AI anxiety to increase workers’ intention to use AI systems.
The model design and definition provide a concrete and fundamental framework for the research and application of the AI system intention to use; adopting the PLS-SEM-ANN methodology improves the accuracy and interpretability of the model assumption validation. This method combines PLS-SEM, used for linear hypothesis testing, with ANN, which excels at capturing nonlinear relationships, providing a robust framework that enhances flexibility in complex analyses, theoretical validation and predictive accuracy. The study results provide an empirical basis for hospital administrators to improve the intention of hospital staff to use AI systems.
