This research evaluates the effectiveness of Retrieval Augmented Generation (RAG) models for hazardous materials transportation, addressing reliability concerns in high-stakes environments.

• Compared performance of three fine-tuned generative models: ChatGPT, Vertex AI, and ORNL RAG
• Assessed how well these models handle specialized transportation safety queries
• Demonstrated how RAG approaches can enhance accuracy and reliability for domain-specific applications
• Identified critical considerations for implementing AI in regulated safety environments

For security professionals, this research provides valuable insights on deploying LLMs in contexts where accuracy is critical and regulatory compliance is mandatory.

Evaluating Retrieval Augmented Generative Models for Document Queries in Transportation Safety