The Cognitive Divide in the Age of Artificial Intelligence: Definitional Power and the Governability of Cognitive Infrastructures

This article introduces the concept of the cognitive divide to describe the gap between societies, organizations, or sectors that are able to retain control over the conditions under which cognitive functions are integrated into, used through, and externalized to AI systems, and those in which these processes occur under constraints and in relationships of dependence on exogenous, opaque, and difficult-to-challenge algorithmic architectures. This divide should not be conflated either with the digital divide or with a simple deficit of individual skills. Rather, it refers to differentiated institutional, productive, and political configurations that determine the capacity to control and govern the cognitive infrastructures that structure perception, evaluation, and collective action. It manifests itself in particular through an unequal distribution of definitional power over the categories, metrics, and rules embedded within these infrastructures. The article argues that responses centered on technological adoption, training, or individual adaptation are structurally insufficient, and that the absence of robust collective mechanisms of coordination, regulation, and productive capacity contributes to the erosion of collective control over cognitive infrastructures. The framework distinguishes three dimensions of control—appropriation, governance, and productive capacity—and develops three ideal-typical AI deployment configurations characterized by different distributions of definitional power, lock-in, and reversibility. Because these configurations differ in the leverage available for intervention, they imply distinct governance priorities rather than a uniform regulatory response.