Redistributing the Bounds of Rationality: How AI Reshapes Decision-Making

Sammanfattning

Decision-making rarely occurs under conditions of full information, unlimited time, or unconstrained cognitive capacity. Herbert Simon’s concept of bounded rationality frames decision-making as a constrained process in which decision-makers (DMs) rely on limited search, heuristics, and aspiration levels to reach satisficing rather than optimal solutions. The rise of artificial intelligence (AI), and generative AI in particular, may relax some of these constraints while introducing new ones such as automation bias. This thesis examines how, if at all, AI adoption reshapes the boundaries of human rationality, and how any such shifts manifest in individual and firm decision-making. The question is addressed through an in-depth qualitative case study of a small e-commerce and fintech firm headquartered in Jordan, drawing on two rounds of semi-structured interviews with nine employees and one external consultant, analyzed thematically against the bounded rationality framework. The findings indicate that AI does not simply expand the bounds of rationality but redistributes the cognitive constraints within them. Search constraints loosen, yet cognitive effort shifts from solution search to problem definition; aspiration levels rise across roles; and heuristics persist in a new form, now governing how far DMs trust AI output rather than which alternatives they generate. Underlying these shifts is a recurring concern that sustained reliance on AI may erode the expertise on which DMs depend.