Digital twin (DT) is increasingly adopted in the architecture, engineering, and construction (AEC) sector to support data-driven modeling, monitoring, and decision-making across the building lifecycle. While Artificial Intelligence (AI) applications in DT have increased rapidly, existing studies remain fragmented, largely task-specific, and weakly integrated into holistic DT frameworks. A comprehensive analysis of the current applications, challenges, and future opportunities is vital for advancing the use of DT in this domain. Guided by three key research questions, this study employs a science mapping methodology to review the application of AI in DT, analyzing 316 publications from 2015 to 2025. The research highlights the most influential journals, countries, and authors, and through a keyword co-occurrence analysis, identifies five emerging research areas. The paper discusses the current implementations and limitations of AI within the realm of DT and offers recommendations for future development. By clarifying the evolving role of AI in DT development, this study provides an up-to-date synthesis of the field and offers structured insights to support the advancement of intelligent, scalable DT applications in the built environment.
Wang et al. (Mon,) studied this question.