Investigating the Potential of ChatGPT in Construction Management: A Study of Interpreting Construction Crane-Related Accident Reports

Large language models, such as Chat Generative Pre-training Transformer (ChatGPT), have shown great potential to revolutionize the construction industry by automating repetitive and time-consuming tasks. This study presents a framework to explore the feasibility of ChatGPT to automatically analyze causal factors using narrative accident reports in construction. Firstly, the ChatGPT was used to extract primary causal factors from accident reports from the perspectives of human-related, crane-related, environment-related, and management-related. Then, outputs from ChatGPT were evaluated by a group of participants in terms of clarity, specificity, reliability, and inspiration. Finally, the outputs from ChatGPT were further corrected by the same participants through careful addition, deletion, or re-judgment of the factors. The findings of this study include: (1) ChatGPT was generally considered a satisfying assisting tool for causal factor extraction and analysis; (2) the results also implied that this technology still needs further development before it can be widely applied in professional areas, such as construction crane safety. Overall, this study highlights the potential and the required further efforts of applying large language models in research.