Transforming transactive memory systems in human-GenAI collaboration

In this study, we examine how generative AI chatbots change knowledge-intensive work by taking the theoretical lens of transactive memory systems (TMS). TMS theory describes how teams coordinate knowledge encoding, storage, as well as retrieval and create a distributed knowledge repository. We argue that generative AI and humans form a hybrid team, in which they contribute to knowledge systems via transforming their specialization, coordination, and credibility. We collected data on open-source software development from GitHub and integrated this field data with a large-scale survey. The empirical results demonstrate that human–AI teams increase their knowledge breadth and depth, thereby enhancing their specialization. For knowledge coordination, no evidence was found for significant changes. Our results for credibility suggest that AI users get assigned more reviews and become more lenient towards their peers’ work. Overall, this study uncovers how human–AI teams contribute and coordinate knowledge and how this impacts existing memory systems in open-source software development.