Assessing the impact of AI-enabled BIM digital twins on construction risks: a Bayesian Network of Jordanian expert...

Building projects in Jordan routinely experience substantial schedule delays and cost overruns, yet robust empirical evidence on the impacts of Artificial Intelligence (AI) and Building Information Modelling (BIM)-based Digital Twins remains scarce. This study develops a belief-based, probabilistic causal model of how AI-enabled BIM Digital Twin technologies may affect time-cost risks in Jordanian reinforced-concrete building projects. Using structured expert elicitation with 600 experienced practitioners (contractors, consultants, project managers, and academics), we capture expert judgements on causal links between four technology clusters-AI-based progress detection, contract-aware BIM assistants, predictive cost-schedule twins, and AI-assisted design/constructability-and key project mechanisms and outcomes. All quantitative results in this study come from these expert judgements; no project-level before-after performance data are used. Accordingly, results are reported as pooled expert beliefs under stated scenarios, not empirically validated predictions. Hierarchical Bayesian models pool elicited probabilities into a Bayesian Network (BN) representing collective expert beliefs, rather than observed project performance. Monte Carlo simulations, Global Sensitivity Analysis (GSA), and Expected Value of Perfect Information (EVPI) are then used to explore technology-adoption and policy scenarios and to rank influential levers. Experts believe that, under idealized, high-readiness conditions, AI-enabled BIM Digital Twins could sharply reduce the probabilities of schedule delays >20% and cost overruns >10%. These figures are belief-based, optimistic upper-bound expectations derived from expert judgement, not empirically measured or validated impacts. The Bayesian Network therefore serves as an uncertainty-aware decision-support tool for early policy and investment decisions in data-poor contexts. It highlights the roles of training, organizational readiness, and client demand alongside technology adoption.