Abstract

Data science has become a high-status information systems (IS) occupation that co-produces expertise with domain experts in AI-enabled decisions, yet formal theories rarely model data scientists as distinct agents. We build an agent-based model extending the garbage can model by introducing data scientists—who pursue data-induced rationality—alongside domain experts—who enact procedural rationality—and by incorporating generative AI (GenAI) as a mediator. Across four phases, we examine how heterogeneous rationalities shape couplings among problems, solutions, and choice opportunities. Three results follow: (1) adding data scientists shifts decision distributions, reducing resolutions while increasing oversight and flight; (2) decoupling roles and rationalities boosts novelty but heightens coordination burdens that depress effectiveness under misalignment; and (3) GenAI improves convergence at moderate—though not extreme—diversity. This model helps us theorize data science as an IS occupation that transforms organizational decision-making, and, more broadly, contributes to the research on AI and decision-making.

Download

Share

COinS