Abstract for: Integrating Artificial Intelligence Techniques into System Dynamics: Opportunities and Challenges on the Path Forward

The integration of artificial intelligence (AI) techniques into system dynamics (SD) may show promising potential for enhancing the dynamic modelling process, structure and behaviour analysis, and decision-making capabilities in complex systems. The current applications of AI to SD rising, but there are still some important fields where AI can excel SD modelling and simulation in the short future. This article explores the advancements and future directions in combining AI with SD modelling. Especially, we highlight specific applications of AI techniques that can be employed at different stages of the SD modelling process, including augmented SD with computer vision, dynamic model building with natural language, and AI generated learning environment design. In our study, we show concrete cases on how to use large language models (LLMs) to create "almost" ready-to-run SD simulation models based on static stock and flow diagrams (SFD). Besides, we also discussed the possible paths towards next-generation SD development. It is expected that AI tools like ChatGPT can play a key role of copilot to help develop model structures and understand dynamic behaviours. Furthermore, we discuss the key benefits, challenges, and opportunities that may arise from AI-enriched modelling and simulation, emphasizing the need for interdisciplinary collaboration and bold attempts in the SD society. ChatGPT aids the development of codes attached in Appendix.