Abstract for: Open Source Tools for AI Supported System Dynamics Modeling & Analysis

Artificial intelligence (AI) is rapidly entering system dynamics model development, promising automation, accessibility, and speed. Yet its use raises fundamental concerns about methodological rigor, transparency, and the preservation of systems thinking skills. This presentation examines all of the open source AI tools available to modelers in the sd-ai platform. The presentation shows a human-in-the-loop, modular AI architecture allowing modelers to embed AI-enabled tools into existing workflows. By creating specialized engines for formulation vs. explanation & critique, the approach emphasizes transparency, validation, and iterative refinement. This design shifts practitioners toward higher-value interpretation and validation while preserving rigor, reproducibility, and systems thinking competencies. This presentation shows how sd-ai enables AI to be used in multiple phases of system dynamics practice. Explanation and narrative engines produce consistent, interpretable insights, while model building engines enable rapid iterative prototyping. These tools show the nascent potential for improved accessibility, faster model exploration, and enhanced learning, without fully displacing expert judgment. Performance varies across engines, underscoring the importance of benchmarking and guardrails. The results clarify that AI’s value in system dynamics today depends strongly on how narrowly and explicitly each tool’s scope is defined. Engines that target bounded tasks, such as explanation, feedback loop narration, or variable documentation perform more reliably than general model generators.