Can a Large Language Model Generate Plausible Business Cases from Agent-Based Simulation Results?

This paper describes new applications of a Large Language Model (LLM) for business domains. So far, we have conducted research on agent-based simulation models to uncover complex socio-technical systems. However, to let ordinary business people understand the models and their consequences, conventional validation or visualization methods are not enough. We must explain the plausible results through cases with natural languages. In our previous studies, we have reported a method for describing simulation results in natural language and grounding them with actual business case. Based on the results, we utilize a Large Language Model for the generation. From this study we have achieved the following results: (1) simulation results are comprehensively analyzed and systematically classified, and (2) the classification results are used as prompts for with a LLM or ChatGPT, and (3) the LLM generates plausible business cases with natural language. We have confirmed that the generated cases are coincide with previous manual generated explanations and easy to understand for ordinary business people.