Intelligent Governance of Complex Social Systems: A Hybrid AI–System Dynamics Model for the Simulation and Projection of Sustainable Energy Transition

Abstract

Amidst growing uncertainty regarding global sustainable development goals and the increasing complexity of social systems, accurately characterizing long-term transition processes driven by the interactions of multiple agents and factors remains a critical challenge. Using China’s energy transition toward carbon neutrality as a proof-of-concept case, this paper constructs a bidirectional coupling framework based on a hybrid Artificial Intelligence–System Dynamics (AI-SD) approach. Within this framework, an axiomatic SD formal system imposes hard constraints on the nonlinear learning processes of the AI, while the nonlinear corrections generated by the AI feed back to modify the system dynamics structure, thereby establishing an iterative cycle of structural constraints, data-driven corrections, and re-imposed structural constraints. Simulation results demonstrate that the hybrid AI-SD model exhibits good comprehensive adaptability in historical fitting and long-term scenario extrapolation, preserving the system feedback structure while characterizing certain nonlinear and time-delay features. The multi-policy synergy scenario, under the conditions set in this paper, shows a high cumulative emission reduction level and exhibits a relatively coordinated characteristic between emission reduction effects and comprehensive social costs. Model-based sensitivity and attribution analyses indicate that, within the defined parameter space, cumulative technology investment and energy cost conditions are associated with higher model output sensitivity; the contribution of carbon pricing shows a certain scenario dependence. This study provides a computable axiomatic framework for the deep integration of mechanistic models and data-driven approaches, and offers theoretical reference for long-term energy governance and complex system decision optimization under deep uncertainty.