Enhancing Preconstruction Education Through Applied Uses of Artificial Intelligence

Abstract

This dissertation examines the intersection of Building Information Modeling (BIM)-based Quantity Takeoff (QTO), preconstruction education, and customized generative AI tools to address the growing need for digital upskilling in construction management (CM), industry skill gaps, and changing expectations for graduate competencies. The study was guided by three related research questions. First, a systematic review was conducted using the PRISMA method to examine the current state of BIM-based QTO research. The review found that BIM-based QTO can improve productivity, accuracy, and completeness in estimating tasks, while challenges remain related to professional skills, CM graduate competencies, and software functionality. Second, the dissertation developed and evaluated DrCGPT, a custom AI teaching assistant created in Microsoft Copilot Studio to support undergraduate students in construction estimating tasks. Student feedback indicated neutral-to-positive experiences and suggested that the AI agent was a useful learning resource, although students continued to value interaction with instructors and peers. Third, the dissertation examined the use of AI agents as primary instructional tools for asynchronous BIM-based QTO learning. Findings indicate that dialogue-tree-based AI agents, supported by controlled knowledge bases, successfully guided students through structured technical tasks and supported self-paced learning, while limitations remained in flexibility and contextual understanding. Overall, this dissertation provides a case-based framework for educators and industry professionals with limited programming experience to responsibly integrate AI agents into teaching and upskilling practices.