Abstract for: Modularization and Integration of Stock-and-Flow Models using ChatPySD

This paper proposes a new method for modular development and integration of System Dynamics models. These modules each embody specific functions, can operate independently, and are less complex than the entire model, making them easier to understand. With the aid of ChatPySD, several modules can be converted into Python modules, and then combined by ChatGPT into a Python application, realizing the complete model function. We have the following requirements for each module: It must have a clearly defined functional scope and accordingly include one or more core stocks. 3. It should be capable of independent simulation within the model development and simulation environment. This means that the module's input parameters should only include stocks and constants, while output parameters can include several other variables in addition to these. This report introduced a method for decomposing existing System Dynamics models into several modules, making them easier to understand. With the aid of ChatPySD, the modules are converted into Python modules, and then recombined using ChatGPT into a Python application that retains the functionality of the complete model. Using two published models as examples, this approach initially demonstrated its feasibility. The purpose of developing this method is to lay the foundation for building a library of System Dynamics modules using the Python language. A higher degree of automation would be desirable. The use of generative AI to support System Dynamics has tremendous potential for development. This approach enables the fusion of human-created and AI-generated system dynamics modules within the same simulation application, thereby advancing the practice of Hybrid Intelligence in the field of System Dynamics. Using ChatPySD to interface ChatGPT's ADA