Explainable artificial intelligence in construction engineering and management: Review of applications, trends, and opportunities

Artificial intelligence (AI) is increasingly used in construction engineering and management (CEM), but limited transparency can restrict trust, validation, and adoption in high stakes decisions. Explainable AI (XAI) can improve transparency, yet explanation methods vary across model types, data structures, and reviewed CEM tasks. This PRISMA guided review examines XAI in CEM journal literature and uses broader current civil engineering publications as bibliometric context. Fifty-five CEM journal articles from Scopus, Web of Science, IEEE Xplore, and ASCE Library were retained for synthesis. Using a taxonomy framework supported by text analysis, the study maps publication trends, themes, application areas, model types, explanation approaches, and limitations. Results show growth in CEM XAI studies, with SHapley Additive exPlanations commonly used to interpret tree based and ensemble models. Key challenges include small and imbalanced datasets, weak validation across contexts, limited human evaluation, and limited integration with building information modeling, digital twins, and automated workflows.