Cognitive Load Analysis in AI-Augmented BI Dashboards: Understanding the Impact of Artificial Intelligence on User Comprehension, Trust, and Decision-Making Efficiency

Abstract

Artificial intelligence integration in business intelligence dashboards fundamentally transforms cognitive processing patterns for decision-makers, challenging traditional assumptions about cognitive load reduction through automation. This article presents comprehensive insights into how AI augmentation redistributes rather than simply reduces mental effort across intrinsic, extraneous, and germane cognitive dimensions. The cognitive landscape becomes significantly more complex when users must simultaneously process traditional data visualizations while interpreting AI-generated insights, explanations, and recommendations. Trust calibration emerges as a critical yet cognitively demanding process, requiring users to continuously evaluate AI system reliability, understand model limitations, and determine appropriate reliance levels on automated recommendations. Evidence reveals that AI-augmented dashboards create new forms of cognitive load, including trust calibration load, explanation processing load, and meta-cognitive monitoring demands that existing theoretical frameworks inadequately address. Individual differences in domain expertise, AI literacy, and automation trust propensity significantly moderate the relationship between AI augmentation and cognitive outcomes. Effective implementation requires human-centered design principles that optimize cognitive load allocation rather than minimizing total mental effort, incorporating layered explanations, progressive complexity training protocols, and adaptive interface features. Organizational adoption strategies must account for heterogeneous user responses, providing tailored support for different trust calibration patterns while monitoring cognitive load indicators alongside traditional performance metrics to ensure optimal human-AI collaborative effectiveness.