Leveraging Generative AI in Enterprise Contexts: Towards a Paradox Theory and Organizational Boundary Work Approach

Firms seeking to implement generative artificial solutions (GenAI) encounter several tensions, such as sharing data to improve GenAI model performance while retaining control over data. These tensions are paradoxical, as they are persistent and require continuous management rather than resolution. This study investigates these paradoxical tensions in the GenAI enterprise context through an exploratory qualitative study, based on interviews with 13 GenAI experts and analyses of secondary data. We construct a two-part theoretical lens to analyze our insights: paradox theory to understand the tensions, and boundary work theory to explore firms’ responses to these tensions. Our preliminary findings identify three main elements – performance, convenience, and control – that explain these paradoxical tensions. Additionally, our results reveal that the strategies for responding to these tensions are intricately linked across various GenAI activities and firms within the GenAI ecosystem. Our research contributes to organizational AI/GenAI, paradox theory, and boundary work literature.