This paper theorizes how language-based artificial intelligence (AI), especially large language models (LLM), may shape the nature of organizational knowing and particularly knowledge integration. Drawing on the concept of language games, we articulate what is conceptually novel about recent advances in language-based AI. Language-based AI can emulate context-aware linguistic practices and thereby translate and apply knowledge across linguistic communities. We elaborate how these new capabilities facilitate de-personification of knowledge and effective translation of knowledge across expert groups. We link these two processes to the contraction of organizational boundaries, the polarization of knowledge work between specialists and generalists, dynamics of innovation, and the erosion of traditional knowledge integration mechanisms, such as shared organizational identity and social networks. We conclude by discussing the broader implications of our arguments for the study of language-based AI technologies, organization-level impacts of AI, and the knowledge-based view of the firm. Overall, our paper seeks to draw attention to and facilitate the analysis of the broader organizational implications of AI beyond its effects on the individual knowledge worker.
