ARTIFICIAL INTELLIGENCE AND MACHINE LEARNING APPLICATIONS IN CONSTRUCTION PROJECT MANAGEMENT: ENHANCING SCHEDULING, COST ESTIMATION, AND RISK MITIGATION

Abstract

Chronic schedule slippage, cost overruns, and safety incidents remain persistent in construction projects, yet empirical, project-level evidence connecting artificial intelligence and machine learning (AI/ML) adoption to core delivery outcomes is limited. This study’s purpose is to quantify how AI/ML adoption relates to schedule adherence, cost estimation accuracy, and risk mitigation effectiveness. We apply a quantitative, cross-sectional, multi-case design using enterprise-scale construction cases sampled across multiple firms. The sample comprises 102 projects meeting strict inclusion criteria for verifiable baselines, actuals, and risk registers. Key variables include an AI/ML Adoption Index 0–100 capturing use-case breadth, run frequency, model sophistication, and integration depth; outcomes are Schedule Delay percent, Cost Overrun percent, Risk Detection Rate, and Residual Risk Impact on a 1–5 scale; a Complexity Index and Digital Maturity serve as moderator and control alongside budget, duration, sector, contract type, and region. The analysis plan proceeds from descriptives and correlations to multiple regression with heteroskedasticity-consistent errors for continuous outcomes, fractional logit for detection rate, interaction testing for Adoption by Complexity, diagnostics, and robustness checks including alternative index weights and fixed effects. Headline findings indicate that each 10-point rise in adoption associates with approximately 1.2 percentage points lower schedule delay, 0.9 percentage points lower cost overrun, 0.04 higher detection rate, and 0.06 lower residual severity, with effects amplified on higher-complexity projects. A structured literature review informed construct operationalization and measurement, synthesizing 47 peer-reviewed studies. Implications for owners, contractors, and CISOs emphasize pipeline-first practices integration to BIM, ERP, and scheduling systems, cadence of model refresh and review, and API-level interoperability to translate analytical signals into routine planning and control.