The rapid development of generative AI (GenAI) tools has ignited debates about their dual role in organizational innovation, capable of enhancing creativity and efficiency while introducing risks like instability, randomness, and hallucinations. Interestingly, the innovation process in organizations is not only a linear and planned procedure but also accommodates variation and randomness as the building blocks of serendipitous discoveries. Drawing on evolutionary theories of innovation, specifically the blind variation and selective retention framework, this study explores how GenAI-induced randomness can foster both deliberate and serendipitous innovations. By synthesizing the literature on AI, digital innovation, and serendipity in management, we have developed a set of propositions and research questions for leveraging AI-driven variation and ensuring responsible AI use in the selective retention of innovations. Specifically, we propose that incorporating GenAI into innovation processes increases the chances of encountering both non-random and random forms of variation, but firms must keep in mind the implicit impacts of GenAI on selective retention of innovative ideas. Our conceptualization advances the understanding of GenAI's impact on innovation processes, offering actionable insights for organizations aiming to harness innovation in dynamic environments.
