Evaluating AI-Powered Building Cost Estimation Accuracy in Quantity Surveying

Thesis Abstract

Accurate building cost estimation is critical for effective project planning and financial management within the construction industry, yet traditional methods often suffer from inaccuracies and inefficiencies, leading to cost overruns and project delays. The integration of artificial intelligence (AI) into quantity surveying practices offers promising potential to enhance the precision and efficiency of cost estimates; however, systematic evaluation of AI-powered estimation tools remains limited. This study aims to evaluate the accuracy of AI-driven building cost estimation systems relative to conventional methods, with a focus on identifying factors influencing their performance and assessing their practical applicability within diverse construction contexts. The specific objectives are to compare the accuracy levels of AI-based estimations with traditional techniques; to examine the influence of project-specific variables on AI estimation accuracy; to develop a predictive model outlining the key determinants of AI system performance; and to evaluate the perceptions of quantity surveyors regarding the usability and reliability of AI tools. A descriptive comparative research design will be adopted, incorporating both quantitative and qualitative data collection methods to ensure a comprehensive analysis. The target population comprises professional quantity surveyors, project managers, and construction cost consultants actively engaged in cost estimation within a metropolitan region with a robust construction industry. A stratified random sampling technique will be employed to select a sample of 150 participants, ensuring representation across different organizational sizes and project types. Data will be collected through structured questionnaires to gauge perceptions of AI tool usability and reliability, as well as semi-structured interviews for deeper insights; additionally, a dataset of 300 past cost estimates generated via both traditional and AI-based methods will be compiled for quantitative analysis. The study will utilize statistical techniques such as paired t-tests and Bland-Altman plots to compare estimation accuracies, supplemented by multivariate regression analysis to identify significant predictors of AI estimation performance. Theories underpinning the research include the Technology Acceptance Model (TAM) to understand user acceptance of AI tools, and the Theory of Reasoned Action (TRA) to interpret behavioral influences on technology adoption. The validity and reliability of research instruments will be established through pilot testing, Cronbach’s alpha, and expert reviews. Data analysis will be conducted using SPSS and NVivo software, with quantitative data analyzed via descriptive and inferential statistics, and qualitative data through thematic analysis to extract key themes from interview transcripts. Expected findings include a statistically significant improvement in estimation accuracy with AI systems compared to conventional methods, particularly in projects with high variability in scope and complexity. Additionally, the study predicts that factors such as data quality, algorithm transparency, and user training significantly influence AI estimation performance. Perceptions from industry practitioners are anticipated to reveal both optimism regarding AI capabilities and concerns related to trust, interpretability, and integration into existing workflows. The study aims to contribute to the theoretical understanding of technological innovation adoption in quantity surveying by extending existing models with context-specific variables, and to offer practical insights for industry stakeholders seeking to leverage AI for cost estimation. The main conclusion underscores the potential of AI to enhance accuracy and efficiency in building cost estimation, provided that critical factors influencing performance are adequately managed. Recommendations include developing standardized protocols for AI system deployment, improving user training programs, and fostering industry-academic collaborations to refine AI algorithms based on real-world data. The study also advocates for further longitudinal research to monitor evolving AI capabilities and their long-term impact on quantity surveying practices, ultimately guiding policy formulation and strategic decision-making in construction cost management.