From AI adaptation to innovation: how leader AI crafting shapes employee innovative behavior

As artificial intelligence reshapes work, organizations face the challenge of helping employees move beyond routine technological adaptation toward innovation. Although prior research has shown that leader AI crafting facilitates employees' adaptation to AI, whether and how it promotes employees' innovative behavior remains unclear. Drawing on social cognitive theory, this study examines how leader AI crafting relates to employees' innovative behavior through AI self-efficacy and when this relationship becomes stronger. We conducted two studies. Study 1 used a two-wave field survey to examine the moderated mediation model linking leader AI crafting, AI self-efficacy, AI task dependence, and innovative behavior. Across the two studies, the results indicate that leader AI crafting enhances employees' AI self-efficacy, which in turn is associated with higher innovative behavior. The positive effect of leader AI crafting on AI self-efficacy is stronger when AI task dependence is high, and moderated mediation effect is also supported. Study 2 used a vignette-based experiment to test the first-stage moderating role of AI task dependence in the relationship between leader AI crafting and AI self-efficacy. These findings extend the leader AI crafting literature by clarifying how employees move from adapting to AI toward innovating in AI-enabled work contexts, and by identifying AI self-efficacy and AI task dependence as key psychological and contextual mechanisms. Given the technology-industry sample in Study 1 the broader generalizability of these findings remains to be examined in future research.