Strategising AI innovation: boosting firm productivity through discontinuous jumps between exploration and exploitation

Research on artificial intelligence (AI) has largely emphasised its productivity-enhancing potential, yet far less is known about how firms can strategically manage the tension between exploration and exploitation during the AI innovation process. Drawing on the literature on sequential ambidexterity, this study examines how discontinuous jumps between exploration and exploitation during AI innovation affect firm productivity. Using panel data on Chinese listed firms, we find that such discontinuous jumps significantly enhance firm productivity. This positive effect is stronger in more munificent industries but weaker when competitive intensity is higher. By providing large-scale empirical evidence on the productivity outcomes of discontinuous jumps during AI innovation and identifying their boundary conditions, this study advances understanding of how firms can temporally organize exploration and exploitation in AI innovation and offers practical insights for managing AI innovation more effectively to enhance productivity.