Artificial intelligence at work: A phenomenon-based interdisciplinary review and groundwork for multilevel scholarship

The implications of artificial intelligence (AI) for work are significant and diverse, yet our understanding of its drivers remains siloed. This is partly due to a fragmented understanding of the AI phenomenon, its examination across diverse disciplines, and the contingent nature of its effects. We aim to help address these issues via two objectives. First, we explore the landscape of research by systematically reviewing how organizational science subdisciplines studying AI conceptualize, characterize, and investigate AI at work and then evaluate how this scholarship clarifies and contextualizes the phenomenon. By examining indicators of these dimensions, we identify distinct clusters of research that represent what we label as "application-orientation" and "generalized-orientation" categories. Comparatively, application-orientation research was the most likely to either define AI’s capabilities concretely or to situate their assessments within a specific function or industry, was less likely to characterize AI as a radically or wholly new and disruptive technology, less likely to contain claims regarding widespread technological unemployment resulting from AI, and less likely to focus on the negative (compared to the positive) outcomes of AI use for workers. Comparatively, generalized-orientation research was less likely to reference AI’s concrete capabilities or situate their analyses in a specific industry context, tended to be less empirical, and was more likely to position AI as radically disruptive or to focus on negative worker outcomes. Second, we seek to add to this research landscape by proposing an illustrative, interdisciplinary multilevel framework that suggests pathways toward balanced, multilevel assessments of the phenomenon.