Abstract

The adoption of artificial intelligence (AI) has improved the efficiency and quality of public services, but also reshaped discretion in public service delivery. Therefore, it is important to examine how AI adoption influences the discretion of street-level bureaucrats. Drawing on a socio-technical systems perspective within a public healthcare context, we investigate how AI adoption influences physicians’ perceived discretion through the interaction of the nature of AI, the behavior of physicians, the attitude of patients, and the organizational environment. Using fuzzy-set Qualitative Comparative Analysis (fsQCA) across 22 departments in the Shandong Provincial Hospital Group, we identify three paths through which discretion is reshaped by adopting AI: (1) the high physician-engagement enforcing path, (2) the high patient-expectation curtailing path, and (3) the high AI-risk curtailing path. Furthermore, our findings show that the effect of AI adoption varies across healthcare contexts, such as between different departments in public hospitals. Our findings offer both theoretical contributions and practical insights into how AI and its social context interact to influence discretion in the public sector.