Causal Inference to Enhance AI Trustworthiness in Environmental Decision-Making

We present a causal model specifically designed for an environmental decision-making context, focusing on agricultural choices and the delicate balance between changes in water policy and agrarian profit. Our comprehensive causal framework incorporates various criteria, including trust sensitivity, which explains the interaction between trust and policy change, actor-specific factors such as location, AI capability, and community awareness of these factors. Additionally, we introduce “trustworthy acceptance” as a metric to measure decision-making progress. The results show that implementing community awareness interventions can increase actors’ trust by up to 21%, thereby reducing the decline in trustworthy acceptance even when profits decrease. By constructing this causal model for environmental decision-making with trade-offs, we can provide a nuanced measurement of outcomes, thus illuminating the complex dynamics of such scenarios.