QUANTITATIVE ASSESSMENT OF AI-ENABLED CONSTRUCTION PLANNING TOOLS FOR REDUCING DELAYS IN U.S. INFRASTRUCTURE PROJECTS

Abstract

This study addresses schedule overruns in U.S. infrastructure projects and the limited quantitative evidence on whether AI-enabled construction planning tools reduce delays. The purpose is to quantify relationships between AI-based planning adoption and schedule performance using project-level data. A quantitative cross-sectional, case-based design used a Likert five-point survey of practitioners covering 198 infrastructure cases across public agencies and enterprise contractors and consultants. Key variables included AI-enabled planning tool adoption, planning quality, coordination effectiveness, project size, complexity, contract type, and a Schedule Delay Index (SDI) from planned and actual durations. Reliability was high for all multi-item scales (α = 0.84-0.91). The analysis plan combined descriptive statistics, Pearson correlations, and multiple regression with moderation tests. Projects showed moderate AI adoption (M = 3.47, SD = 0.78) and an average 11% schedule overrun (SDI M = 0.11, SD = 0.09). AI adoption correlated negatively with SDI (r = −0.41, p < .001) and remained a significant predictor of lower delay after controlling for size, complexity, and contract type; a one-point increase in adoption was associated with a 2.8 percentage point reduction in SDI. Adding planning quality and coordination effectiveness increased explained variance in SDI from 25% to 41% and partially mediated the AI-delay relationship, with effects strongest on highly complex projects. The headline finding is that AI-enabled planning tools contribute meaningfully to delay reduction when embedded in robust planning and coordination practices. The study implies that infrastructure owners should treat AI-enhanced planning as a strategic capability for improving delivery reliability across the sample.