The impact of AI technology application on street-level bureaucratic decision-making: A moderated mediation experimental analysis

The integration of Artificial Intelligence (AI) into public administration reshapes the decision-making environment for street-level bureaucrats. Against this backdrop, this study investigates how their trust of AI-assisted decisions is influenced by cognitive state, algorithmic interpretability, and organizational culture. Furthermore, it examines the mediating role of risk perception and the moderating effect of AI literacy as key influencing conditions. Through a scenario-based survey experiment with 664 street-level bureaucrats, this study finds that cognitive, algorithmic, and organizational factors all significantly influence perceived trust in AI decisions. Specifically, the influence of cognitive conflict and algorithmic interpretability on decision adoption is mediated by risk perception; however, this mediating path does not hold for organizational culture. Additionally, a bureaucrat's AI literacy mitigates the perceived risk associated with low-interpretability algorithms, though it does not alter the effect of their cognitive state. By exploring the complex mechanisms of AI acceptance in public service, this study offers significant implications for designing human-centric AI and fostering effective human-AI collaboration.