Explainable artificial-intelligence methods in decision support for delivery management

The article presents an analysis of the application of explainable artificial intelligence (XAI) methods in supporting managerial decision-making for delivery management. The study is conducted within a theoretical and analytical framework, employing a targeted review of publications, a taxonomy of interpretable methods, and their alignment with typical delivery tasks. Particular attention is paid to the role of XAI in ensuring transparency, consistency, and resilience of decisions, as well as integrating ESG indicators and cyber-resilience criteria into planning and dispatching processes. Based on the analysis of scientific sources, approaches to interpreting forecasts and recommendations (post-hoc methods, built-in relevance mechanisms, neuro-symbolic models) are systematized, and their managerial effects and limitations are identified. Generalizing classifications are presented, including a map of XAI applications in logistics tasks, a summary of quantitative effects by key metrics, and a decision matrix linking approaches to managerial criteria. The challenges of XAI implementation are examined, including the complexity of interpreting deep models, the risk of oversimplified explanations, and the need to adapt algorithms to dynamic operational conditions. A structured model for integrating XAI into decision support system architectures is proposed, focusing on balancing accuracy, cost, and environmental performance. The article will be useful to specialists in logistics and supply chains, DSS developers, XAI researchers, and managers making decisions under high demands for transparency and delivery resilience.