Abstract
Artificial intelligence (AI) is rapidly becoming a primary channel through which employees access information and solve problems at work. Drawing on media dependency theory, we theorize that workplace AI use exerts a suppressing effect on practitioners’ job self-efficacy in task-focused achievement contexts. Specifically, AI use can directly bolster job self-efficacy by facilitating task accomplishment. At the same time, frequent AI use may foster AI dependency, reducing employees’ engagement in competence-relevant cognitive work and, in turn, weakening mastery-based confidence. Across a multi-source, time-lagged field study and a scenario experiment, results support this competing-process account: workplace AI use showed a positive direct association with job self-efficacy, whereas AI dependency carried a negative indirect effect, yielding a significant suppressing effect. We further identify proactive feedback to AI as a key boundary condition that attenuates the negative effect of AI dependency on job self-efficacy and renders the suppressing effect nonsignificant at higher levels of feedback. This research clarifies how organizations can leverage AI for performance gains while safeguarding employees’ perceived capability.