ChatPySD: Embedding and Simulating System Dynamics Models in ChatGPT‐4

This paper presents a novel approach for seamlessly integrating system dynamics models into the ChatGPT‐4 conversational AI. By transforming system dynamics models with PySD and embedding them into ChatGPT's conversational context, users can explore even complex models through natural language interactions. ChatGPT‐4 is capable of running simulations based on user‐provided parameters and offering explanations and analyses of the simulation results, leveraging its extensive knowledge base. Because of the constraints of ChatGPT's analytical and computational capacity, it is crucial to provide additional essential information alongside the system dynamics model and to thoroughly verify and implement ChatGPT's responses. ChatGPT, in turn, can dynamically generate executable Python code, creating a bidirectional interface that empowers modelers to engage with ChatGPT using programming languages. This complements existing natural language and image‐based interactions, enhancing the overall efficiency of communication. Successfully integrating system dynamics models with ChatGPT not only makes the models more accessible but also enhances various aspects of system dynamics, including modeling, simulation, discussion, and education. Furthermore, the inherent rigor and transparency of system dynamics models can be extended to ChatGPT conversations, thereby expanding ChatGPT's capabilities.