How AI usage shapes employees’ exploitative and exploratory innovation: joy as a mediator and learning goal orientation as a moderator

Introduction Employees play a pivotal role in organizational ambidextrous innovation, yet existing studies have paid limited attention to how artificial intelligence shapes employees’ exploitative and exploratory innovation. Drawing on cognitive appraisal theory and the broaden-and-build theory of positive emotions, this study identifies joy as the central emotional mechanism linking artificial intelligence usage to these two forms of innovation. As a high-arousal positive emotion grounded in person–situation fit, joy promotes active engagement, in contrast to lower-arousal emotions such as satisfaction or happiness that reflect acceptance rather than pursuit. We further examine how learning goal orientation moderates the extent to which joy translates into exploitative and exploratory innovation, thereby advancing understanding of how technological empowerment affects ambidextrous innovation. Methods This study draws on survey data from Chinese employees ( N = 669) and employs partial least squares structural equation modeling (PLS-SEM) to examine the mediating role of joy in the relationship between artificial intelligence usage and employees’ exploitative and exploratory innovation, as well as the moderating effect of learning goal orientation. Results Artificial intelligence usage shows small but meaningful positive effects on employee exploitative innovation ( β = 0.120) and exploratory innovation ( β = 0.104), with joy partially mediating the effect on exploratory innovation only (indirect β = 0.050). Moreover, joy positively predicts exploitative innovation ( β = 0.182) and exploratory innovation ( β = 0.206) only under high learning goal orientation. Discussion The findings emphasize that the role of positive emotions is not universal but rather motivation-dependent: joy mediates exploratory but not exploitative innovation, while high learning goal orientation amplifies its effects on both innovation types. These results extend emotion theories to AI-enabled work contexts and offer practical implications for fostering employees’ emotions and learning motivation to achieve synergy between technological empowerment and innovation.