Unveiling diverse AI applications in BIM: a quantitative mapping of techniques for 10 dimensions
Samereh Jadidoleslami, Ehsan Saghatforoush
Abstract
The integration of Artificial Intelligence (AI) and Building Information Modeling (BIM) has attracted growing research interest, though studies remain concentrated on lower BIM dimensions, with limited attention to higher ones such as 9D (lean construction) and 10D (industrialized construction). This study systematically reviews 122 peer-reviewed articles to map AI applications across all ten BIM dimensions and project lifecycle phases using bibliometric, quantitative, and thematic analyses. Excel, VOSviewer, and NVivo were used for numerical, network, and qualitative analyses, respectively. The findings reveal that Machine Learning (ML) and Deep Learning (DL) are the most frequently used AI methods-appearing 87 and 82 times-mainly in 3D (72.1%) and 6D (23%) dimensions related to geometric modeling and sustainability. In contrast, 4D (time scheduling), 9D (lean), and 10D (industrialized construction) are critically underexplored. Lifecycle analysis shows the dominance of design and construction phases, while planning, maintenance, and end-of-life receive minimal focus. The study offers a comprehensive mapping of AI-BIM integration, highlights gaps in data-rich and sustainability-oriented areas, and suggests future research directions, such as integrating audio-visual data, transfer learning, and NLP or computer vision techniques to advance higher-dimensional and underrepresented lifecycle applications.