Work

Information Overload and Disutility in Generative AI Systems as Cognitive Extension: A Bounded Rationality Perspective

Year: 2026
Type: preprint
Abstract: Generative AI systems promise cognitive extensions by enhancing information access and use. However, from the bounded rationality perspective, their utility is fundamentally constrained by inherent human cognitive limitations. This article argues that when the volume of AI-generated information exceeds users’ processing capacity governed by working memory constraints, attentional resources, and verification capabilities, it produces cognitive disutility rather than enhancement. We introduced a formal model to quantify this interaction, demonstrating that net cognitive benefits decline systematically as information overload increases beyond individual processing thresholds. The analysis reveals that successful cognitive extension depends not on maximizing informational output but on strategically aligning AI design with human cognitive ergonomics and individual differences in processing capacity. Ultimately, truly augmentative systems must minimize cognitive overhead to prevent overload, while preserving users' epistemic agency and decision-making autonomy.
Language: English
Open Access status: green
OpenAlex ID: W7236217147
Cites: 41
Cited by:
Related to: 10