Explainable artificial intelligence (XAI): ensuring transparency and user trust in educational decisions — a conceptual model

Abstract

This article examines the application of Explainable Artificial Intelligence (XAI) methods in educational systems to ensure transparency of AI-driven decisions and foster trust among students and educators. The conceptual foundations of XAI are systematized — including ante-hoc and post-hoc interpretation approaches, key methods (SHAP, LIME, Attention Visualization, ANCHORS), and explainability taxonomies. A four-level conceptual model of user trust in educational AI systems is developed, encompassing data, model, interface, and governance layers. A comparative analysis of XAI methods is conducted based on transparency, explanation accuracy, and educational applicability criteria. The need for XAI systems to comply with EU AI Act 2024 regulatory requirements is justified. Recommendations for integrating XAI into AI educational platforms at Uzbek higher education institutions are provided.