Explainable Artificial Intelligence for Strategic Decision-Making in Management Information Systems: A Critical Review of Transparency, Trust, and Governance Frameworks

Abstract

Artificial intelligence is increasingly embedded in management information systems to support strategic decisions involving forecasting, resource allocation, market intelligence, risk assessment, supply-chain resilience, financial control, cybersecurity, human-resource analytics and organizational governance. Strategic decision-making differs from routine operational automation because it involves uncertainty, long-term consequences, value-laden trade-offs, reputational exposure, regulatory obligations and human accountability. In such settings, predictive accuracy alone is insufficient. Managers require explanations that clarify why an AI system recommends a course of action, what evidence and assumptions shaped the recommendation, how reliable the output is under changing conditions and who remains accountable when algorithmic advice influences organizational outcomes. Explainable artificial intelligence (XAI) has therefore become a critical capability for transforming black-box machine-learning outputs into decision-relevant, contestable and auditable knowledge. This review critically examines XAI for strategic decision-making in management information systems, focusing on transparency, trust calibration and governance frameworks. The paper synthesizes interpretable modelling, post-hoc local explanations, feature attribution, counterfactual explanations, surrogate models, causal explanation and human-centred explanation interfaces. It further evaluates how these approaches influence managerial trust, decision quality, accountability, compliance and organizational learning. The review argues that XAI should not be treated as a technical add-on to predictive modelling; it must be embedded across the complete decision lifecycle through data governance, model documentation, explanation quality controls, stakeholder participation, human oversight and continuous monitoring. A conceptual XAI-GovMIS framework is proposed to connect data governance, model transparency, explanation design, human-AI interaction, strategic decision accountability and responsible AI governance. The paper concludes by identifying unresolved research gaps, including explanation overload, performative transparency, overtrust, weak empirical validation, causal insufficiency, cross-functional accountability gaps and the need for sector-specific governance models for AI-enabled management information systems.