Knowing Who To Ask: Exploring How GenAI Tools Afford and Constrain Expertise Coordination in Software Development Teams

Sammanfattning

Generative AI (GenAI) tools have seen rapid adoption in software development teams, yet research has focused almost exclusively on individual-level outcomes such as productivity and task completion speed. The consequences for how teams coordinate expertise remain unexplored. This study examines what the affordances and constraints of GenAI tools are for software development teams and how they interact with expertise coordination practices. Drawing on Faraj and Sproull's expertise coordination framework, transactive memory systems theory, and enacted affordance theory, eight semi-structured interviews were conducted with software developers and project managers. The findings confirm that GenAI tools afford considerable individual-level benefits including efficiency gains, expanded problem-solving capacity, and democratized access to technical knowledge. At the same time, the study identifies a set of coordination-level consequences that are structurally invisible to the people experiencing them. The study argues that these affordances and constraints are two dimensions of the same structural problem: GenAI tools operate on a categorically different substrate from the tacit, relational ecosystem on which expertise coordination depends. This finding offers practical implications for software development teams and organizations seeking to understand the long-term effects of GenAI adoption on team knowledge and coordination.