Purpose This study aims to comprehensively examine the integration of artificial intelligence (AI) and building information modeling (BIM) within the building construction field and identify four key enablers to develop PERGE as an AI–BIM adoption framework. It aims to evaluate the applicability of AI methods, including generative methods, and identify emerging trends and underexplored combinations of AI methods and use cases. Design/methodology/approach A scientometric methodology was adopted to establish the AIBI dataset, including 971 peer-reviewed publications, and analyze them based on a computational review and evidence gap maps (CEGMs) approach. A structured query was designed to identify relevant investigations, which were then analyzed to map publication trends, identify dominant AI applications in buildings, perform temporal analysis of recent developments and develop a construct–outcome heatmap. Findings The analysis reveals a significant evolution in AI–BIM integration for buildings, with a shift from early automation tasks to more advanced objectives such as generation, prediction and semantic understanding. There is a notable rise in the use of large language models, reinforcement learning and fine-tuned transformers. The study also identifies a transition in methodological focus from general prediction tasks to the development of algorithmic frameworks tailored to facility needs. Generative AI has notably influenced expectations and applications in the field while also exposing gaps in underutilized areas. Originality/value This paper provides a novel, data-driven synthesis of AI–BIM integration investigations in building construction, energy and facility management, with a particular emphasis on the transformative role of generative AI. The novel adoption framework of PERGE is established, offering valuable insights for researchers and practitioners along with the identification of key trends, suitable AI methods and underexplored opportunities.
Samad M. E. Sepasgozar (Thu,) studied this question.