EXPLAINABLE AI SYSTEMS FOR STRATEGIC ADMINISTRATIVE DECISIONS IN UNIVERSITIES: A STRUCTURAL EQUATION MODELING STUDY

Despite the spread of AI-driven predictive analytics systems in higher education, their adoption in strategic administrative decision-making remains limited, principally due to concerns about algorithmic transparency and trust among academic leaders. While machine learning models prove 87–93% accuracy in predicting student outcomes, implementation gaps keep it up between technological capabilities and actual utilization in administrative contexts. This study examined the causal relationships between AI-driven decision support systems (AI-DSS), explainable AI (XAI) mechanisms, academic leaders' trust, and strategic administrative decision quality in universities, testing the mediating role of algorithmic transparency. The research addresses a critical gap in understanding how transparency features affect the translation of AI capabilities into enhanced governance outcomes. A cross-sectional survey design was employed with data collected from 387 academic leaders across 22 Egyptian universities. Structural equation modeling (SEM) was used to test a hypothesized causal model. Results revealed that algorithmic transparency through XAI mechanisms is not merely a desirable feature but a critical prerequisite for enhancing academic leaders' trust and realizing the potential of AI systems in strategic administrative decision-making. The absence of a direct effect from AI-DSS implementation to decision quality indicates that deploying sophisticated predictive analytics alone is insufficient; transparency features that enable understanding of algorithmic logic are essential mediating mechanisms. Findings provide empirical evidence for institutions investing in AI infrastructure to prioritize explainability features—such as SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-agnostic Explanations)—alongside predictive accuracy. This study offers actionable insights for higher education administrators, EdTech developers, and policymakers navigating AI governance frameworks.