Large language models (LLMs), exemplified by OpenAI’s Generative Pre-trained Transformer
(GPT), represent a new frontier in generative AI. This presentation will provide a brief introduction
to LLM technology and discuss the opportunities it offers to transform pavement engineering. One
such opportunity is improving the understanding of the causes of highway construction accidents.
By leveraging the wealth of publicly available data related to accidents and investigations, previously underutilized textual data can now be explored with the emergence of LLMs. The presentation introduces an approach to utilizing this textual data with a specific focus on the highway sector of the construction industry, which encompasses a wide range of construction activities. OpenAI’s GPT-3.5 LLM was guided to perform summarization and classification across the entire database, yielding unique insights that would have otherwise been difficult to extract from the database alone
