Study of the co-evolution of heterogeneous policy combinations with AI-driven carbon emission monitoring and dynamic carbon trading mechanisms

Artificial intelligence (AI) is reshaping global carbon governance, yet China's current policy framework suffers from fragmentation. To explore policy coordination mechanisms in AI-driven carbon governance, this study constructs a tripartite evolutionary game model involving government, high‑carbon-emission enterprises, and AI monitoring platforms. Intertemporal heterogeneous policies are incorporated. Using China's steel industry as a case, system dynamics simulations evaluate 16 policy scenarios. Results show: (1) Heterogeneous policy combinations exhibit non-linear synergistic effects, with single policies converging toward suboptimal equilibria, failing to incentivize the transition of AI platforms toward precise monitoring strategies. Dual-policy combinations show mixed results — “carbon trading infrastructure + international certification” achieves system transition to (1,1,0.8), while reward-punishment combinations fail due to “subsidy dependency.” The four-policy combination establishes an intertemporal coordination mechanism, achieving optimal equilibrium K8(1,1,1), which significantly enhances monitoring accuracy and carbon market allocation efficiency. (2) The four-policy combination demonstrates structural robustness, with most parameters converging to the ideal equilibrium K8(1,1,1) within ±25% and ± 50% perturbation ranges. However, certain parameters, such as Cah and We, may reverse the system toward the non-equilibrium K(1,1,0) under a 50% perturbation, while the initial policy selection probability has no significant effect on evolutionary outcomes. (3) Positive feedback loop R1 exists between enterprises' proactive emission reduction and AI monitoring platforms' precise monitoring. When dominant mechanisms shift from positive to negative feedback, the system deviates from optimal trajectories, evolving from the ideal equilibrium point K8(1,1,1) toward suboptimal equilibrium points. (4) The system forms five stable equilibrium points, with K8(1,1,1) representing the ideal equilibrium point—a state of comprehensive synergistic governance characterized by active regulation, proactive emission reduction, and precise monitoring.