The climate crisis is becoming a pressing concern for humanity, and policymakers face the challenge of collecting and processing information to formulate context-sensitive policies to mitigate its impact. This study explores the potential of generative agents in developing context-specific climate policies using minimal but strategically relevant data inputs. It considers key environmental indicators as input for five countries: Brazil, China, Germany, India, and the United States. Findings reveal recurring policy themes across countries, such as the necessity of renewable energy expansion and community engagement. Additionally, this study demonstrates the ability of AI-driven approaches to generate actionable policy recommendations tailored to a country's socio-economic and political context. This study further examines the alignment of AI-generated policies with official Nationally Determined Contributions using semantic similarity analysis and also examines the marginal contribution of individual agents through an ablation study. Overall, in the context of the AI-based climate literature, this study shifts the focus from predictive analysis to actionable decision-making and contributes to the growing discourse on the role of generative agents in climate governance by developing context-sensitive climate policies through a multi-agent framework relying on minimal data inputs.
ISBN for host publication: 979-8-3315-4497-3