AI for Global Climate Cooperation: Modeling Global Climate Negotiations, Agreements, and Long-Term Cooperation in RICE-N

Tianyu Zhang, Andrew Robert Williams, Phillip Wozny, Kai-Hendrik Cohrs, Koen Ponse, Marco Jiralerspong, Soham Rajesh Phade, Sunil Srinivasa, Lu Li, Yang Zhang, Prateek Gupta, Erman Acar, Irina Rish, Yoshua Bengio, Stephan Zheng

TL;DR: A climate-economic simulator using multi-agent RL to evaluate global climate negotiation strategies, showing how they can improve climate outcomes and equity.

Abstract: Global cooperation on climate change mitigation is essential to limit temperature increases while supporting long-term, equitable economic growth and sustainable development. Achieving such cooperation among diverse regions, each with different incentives, in a dynamic environment shaped by complex geopolitical and economic factors, without a central authority, is a profoundly challenging game-theoretic problem. This article introduces RICE-N, a multi-region integrated assessment model that simulates the global climate, economy, and climate negotiations and agreements. RICE-N uses multi-agent reinforcement learning (MARL) to encourage agents to develop strategic behaviors based on the environmental dynamics and the actions of the others. We present two negotiation protocols: (1) Bilateral Negotiation, an exemplary protocol and (2) Basic Club, inspired from Climate Clubs and the carbon border adjustment mechanism (Nordhaus, 2015; Comissions, 2022). We compare their impact against a no-negotiation baseline with various mitigation strategies, showing that both protocols significantly reduce temperature growth at the cost of a minor drop in production while ensuring a more equitable distribution of the emission reduction costs.

Lay Summary:

Getting countries to agree on tackling climate change is incredibly difficult. Nations often have conflicting priorities, and there's no global authority to enforce agreements, making it hard to ensure fair and effective action. Many current climate predictions also assume countries will simply follow prescribed plans, without fully exploring if that’s realistic given their individual self-interest.

To address this, we developed a sophisticated computer simulation called RICE-N. In this virtual world, AI-driven "regions" learn to make strategic decisions about their economies and climate policies, mimicking real-world complexities. We then introduced and tested different ways for these regions to negotiate climate agreements, including a novel "Basic Club" approach inspired by real-world proposals.

Our simulation shows that these structured negotiation methods can lead to significant cuts in global warming with only a minor impact on the global economy. Importantly, they also help ensure the costs of fighting climate change are shared more fairly between regions. This research provides a more realistic way to explore how different international negotiation strategies could help us achieve effective and equitable global climate solutions.

Application-Driven Machine Learning: This submission is on Application-Driven Machine Learning.

Link To Code: https://github.com/mila-iqia/climate-cooperation-competition

Primary Area: Applications->Social Sciences

Keywords: reinforcement learning, climate change, mechanism design

Submission Number: 3683