How AI Adoption Shapes Construction Workers’ Extra-Role Safety Behavior: The Mediating Roles of Responsibility and Relational Energy

The integration of artificial intelligence (AI) within the construction industry has not only enhanced safety measures but also increased protection and efficiency for workers. However, viewing AI as “co-workers” rather than simply tools poses fresh challenges to maintaining safe behaviors among workers. This study uses the diffusion of responsibility theory and attribution theory to develop a model that examines the responses of workers when AI is integrated into their work tasks. We hypothesize that higher levels of AI adoption may diminish workers’ sense of safety responsibility and weaken relational energy, which, in turn, adversely affects their extra-role safety behaviors. Using a mixed-methods approach, which blends field studies with experimental design, we collected 425 data sets from a major smart construction site in China. Our analysis in a human–AI collaborative setting revealed significant serial indirect effects of AI adoption on extra-role safety behavior, mediated by changes in safety responsibility and relational energy. Additionally, we found that perceived AI usefulness and worker adaptability enhance these effects. These findings contribute to the literature on AI and safety behavior by identifying sense of responsibility and relational energy as critical psychological mechanisms that link AI adoption to behavioral outcomes. By leveraging diffusion of the responsibility theory and attribution theory, this study elucidates the psychological responses of workers to AI integration, thus enriching theoretical perspectives. Additionally, it provides practical recommendations for managing AI implementations in ways that maintain essential safety standards and promote effective human–AI collaboration in the construction sector.