12814: Knowing the Unknowable With Collective Bricolage: Communities of Practice for Generative AI Tools (session 26395)

Recent advancements in AI technology have led to a rapid expansion of generative AI tools in organizations, enabling the creation of new content, data, or solutions by learning patterns from existing data. However, it remains unclear how users can fully understand the inherent possibilities and most effective use of these tools, as they differ from traditional tools by being fundamentally unknowable. To address this, we explore how an online community engages in collective knowledge creation around generative AI tools. Leveraging previous research on communities of practice, we focus on how communities learn about tools that are intrinsically opaque and contain elements of randomness in their outputs. Our empirical study uses netnography and systematically analyzes forum contributions from the OpenAI Developer Forum over a four-month period. Our findings suggest that knowledge creation regarding generative AI tools follows a process of collective bricolage, where community members engage in three distinct activities: experimentation, accumulation of resources, and dialoguing with and (dis)assembling resources from the shared stock of knowledge. Our study contributes to bricolage and communities of practice literatures and offers a roadmap for effectively developing and disseminating knowledge related to generative AI tools through online communities.

view paper (if available)