The Role of Large Language Models for Decision Support in Fire Safety Planning

This paper explores the integration of Large Language Models (LLMs), specifically GPT-4, in fire safety planning and knowledge-based systems within the Architecture, Engineering, and Construction industry. Focusing on overcoming challenges in expert systems, it presents an AI-in-the-loop model, illustrating how LLMs enhance decision support. The paper introduces a scenario analysis approach, demonstrating the iterative use of LLMs to enrich expert systems' knowledge bases. Two case studies emphasize the practical application of this approach in fire safety planning, showcasing LLM adaptivity, specialized reasoning, and domain knowledge integration. The study addresses the challenges of LLM-induced hallucinations and emphasizes the need for further research to enhance reliability in technical scenarios. Overall, it contributes to advancing fire safety strategies by leveraging the strengths of LLMs in dynamic building environments.