The Rationality Illusion: Why AI-Driven Decision Systems Undermine Organizational Intelligence

Abstract

Organizations increasingly adopt artificial intelligence systems under the assumption that computational efficiency, data-driven consistency, and predictive accuracy translate into superior decision-making. This article challenges that assumption by examining how algorithmic decision systems systematically erode organizational rationality even as they enhance certain computational capabilities. Drawing on bounded rationality theory and extensive empirical research across healthcare, criminal justice, human resources, and public administration, the analysis identifies four interconnected mechanisms through which AI diminishes decision quality: metric displacement (optimizing measurable proxies rather than authentic objectives), cognitive compression (narrowing human judgment around algorithmic defaults), contextual erasure (eliminating situational particularity essential to sound judgment), and reflexive capacity atrophy (suppressing organizations' ability to question their own premises). These mechanisms produce cascading institutional consequences including accountability diffusion, contestability reduction, and adaptive learning deterioration. The findings suggest that AI functions optimally as a bounded computational tool rather than as a rationality substitute, and that organizations treating algorithmic outputs as inherently superior judgment systematically compromise their institutional intelligence.