Human-Centred AI and the Myth of Cognitive Neutrality A Structural Critique of Implicit Human Models in Intelligent Systems

Human-Centred Artificial Intelligence (HCAI) has emerged as a dominant paradigm aimed at aligning intelligent systems with human needs, values, and capabilities. However, this paper argues that many HCAI frameworks implicitly rely on unexamined assumptions about “normal” cognition, leading to forms of systemic exclusion that are neither intentional nor easily detectable. Rather than stemming from ethical neglect, these failures arise from architectural blind spots in how cognitive diversity is represented—or ignored—within system design. This work introduces the concept of cognitive normativity as a critical lens for analysing Human-Centred AI systems. It proposes that without an explicit layer addressing cognitive assumptions, HCAI risks optimising for a narrow subset of human cognitive profiles while presenting itself as universally inclusive. By reframing the problem as a structural limitation rather than a moral one, this paper offers a conceptual foundation for more robust, inclusive, and cognitively plural AI design. The implications of this critique extend across domains including digital work environments, education, accessibility, and intelligent automation, positioning cognitive diversity as a core design variable rather than a peripheral concern.