Extracting FMEA information from publicly available datasets using large language models
Rahul Sharan Renu · Proceedings of the Design Society · 2025
ABSTRACT: The objective of this research is to explore the use of publicly available recall data from the National Highway Transportation Safety Administration to extract Failure Modes and Effects Analysis data. This large data set was analysed using a Large Language Model chatbot. To assess the usefulness of priming the chatbot with this data, the chatbot was also asked to generate data without priming it with the recall data. This was performed on two specific products. It was found that primed-chatbot results were more specific and used technical terminology appropriate to the product being analysed. The proposed approach can be used by designers in the forward design process during new product development. The proposed approach provides designers with insight into potential failures, the associated consequences, their severity, and root causes as well.