When Bots Join the Team: Bot Adoption and the Institutional Fabric of Open-Source Software Projects

This research investigates how the integration of AI agents into human teams affects group organization. By treating bots as active participants rather than mere tools, the authors explore whether the presence of automated agents strengthens or weakens the social structure of open-source software projects. The study focuses on how these digital participants influence coordination, conflict, and the nature of the work produced.

Analyzing Bot Participation

The researchers examined 2,991 GitHub projects, analyzing data from two years before and two years after each project adopted its first bot. To understand how bots impact a team, the study applied institutional theory, measuring three key capabilities: repeated engagement, social memory (the recognition of specific bots in discussions), and role differentiation. The authors then tracked two primary outcomes: the frequency of "conflict cascades" and the distinctiveness of the project's output.

Key Findings

The study found that the adoption of a bot is followed by measurable shifts in team dynamics. Specifically, projects experienced more repeated collaboration, increased recognition of bots within team discussions, a reduction in conflict cascades, and more distinctive project outputs. Notably, these changes did not accumulate gradually over time; instead, they clustered specifically around the moment of bot adoption, suggesting that the arrival of an automated agent serves as a catalyst for organizational change.

The Role of Social Infrastructure

The findings suggest that predictable, rule-based agents can effectively become part of a community's social infrastructure. The data indicates that capabilities—such as coordination and differentiation—predict outcomes based on their function, regardless of whether they are performed by a human or a bot. While human-side capabilities helped explain the reduction in conflict, they did not fully account for the increase in output distinctiveness, suggesting that the bots themselves contribute uniquely to the team's work.

Important Considerations

The authors emphasize that because the study lacks an untreated comparison group, the results should be interpreted as precisely timed associations rather than definitive causal effects. The paper concludes that while the bot acts as the "occasion" for these changes, the underlying mechanism is the evolution of the community's social organization. The bot does not replace human interaction but rather integrates into the existing social fabric to influence how the team coordinates and produces work.