Leadership as a boundary condition on trust in AI: Transformational leadership, leader-member exchange, and employee intentions to use generative AI

Abstract

The onset of generative artificial intelligence (GenAI) has outpaced organizations’ understanding of what makes employees willing to use it. This study asks whether leadership quality shapes how employees translate beliefs about AI into intentions to use it at work. Adapting the multidimensional online trust instrument of Hwang and Lee (2012) to the AI context, we measured uncertainty avoidance, descriptive norms, perceived reliability, and perceived ability, and tested each as a predictor of behavioral intention to use AI. Leadership was assessed twice in the same sample, once as transformational leader behavior (Podsakoff et al., 1990) and once as leader-member exchange (Graen & Uhl-Bien, 1995), allowing the moderation to be tested against two distinct leadership constructs. Working adults recruited through LinkedIn completed the survey (N = 97 for models involving AI measures). Perceived AI ability predicted behavioral intention, and transformational leadership moderated that relationship. The interaction replicated with LMX-7: the association between perceived ability and intention was stronger under higher-quality leadership. Descriptive norms predicted intention but showed no interaction; uncertainty avoidance predicted nothing. Leadership was also associated with significantly lower intention among the roughly one quarter of employees who rated AI’s capability lowest. This paper discusses implications for theory and practice. Perceived reliability, perceived ability, and behavioral intention were each measured with a single adapted item, so these findings are exploratory.