Beyond the Black Box: A Human-Centered Framework for Explainable Artificial Intelligence in Enterprise Decision Making

Abstract

The accelerating deployment of artificial intelligence within enterprise decision environments has intensified scholarly and managerial concern over the opacity of algorithmic reasoning. This paper examines how human-centered explainable artificial intelligence (XAI) can reconcile the analytical power of machine learning with the interpretive, ethical, and accountability requirements of organizational decision making. Employing a qualitative, theory-driven synthesis of peer-reviewed literature spanning computer science, information systems, and organizational behavior, the study integrates technical scholarship on interpretability methods with human-computer interaction research on trust, cognition, and algorithm aversion. The analysis is structured around a conceptual framework that positions explanation not as a terminal technical output but as a relational process linking data, model, explanation, human judgment, and organizational action. Findings indicate that explainability functions as a boundary object between technical and managerial communities, that trust in AI-assisted decisions is calibrated rather than fixed, and that enterprise adoption of XAI is shaped as much by governance structures and workforce readiness as by algorithmic sophistication. The paper further identifies a persistent tension between model accuracy and interpretability, and demonstrates that this tension is frequently overstated in practice when explanation is treated as a design requirement rather than an afterthought. The study contributes an integrative framework for human-centered enterprise XAI, clarifies conceptual ambiguities surrounding interpretability and explainability, and offers guidance for scholars and practitioners seeking to design accountable, trustworthy, and organizationally embedded AI decision systems.