A Trustworthy and Responsible Decision-Making Framework for Resource Management in Food-Energy-Water Nexus: A Control-Theoretical Approach

This paper introduces a hybrid framework for trustworthy and responsible natural resource management, aimed at building bottom-up trust to enhance cooperation among decision makers in the Food, Energy, and Water sectors. Cooperation is highly critical for the adoption and application of resource management alternatives (solutions), including those generated by AI-based recommender systems, in communities due to significant impact of these sectors on the environment and the economic productivity of affected communities. While algorithms can recommend solutions, effectively communicating and gaining community acceptance of these solutions is crucial. Our research stands out by emphasizing the collaboration between humans and machines, which is essential for addressing broader challenges related to climate change and the need for expert trade-off handling in the management of natural resources. To support future decision-making, we propose a successful control-theory model based on previous decision-making and actor behavior. We utilize control theory to depict how community decisions can be affected by how much individuals trust and accept proposed solutions on irrigation water rights and crop operations in an iterative and interactive decision support environment. This model interacts with stakeholders to collect their feedback on the acceptability of solutions, while also examining the influence of consensus levels, trust sensitivities, and the number of decision-making rounds on the acceptance of proposed solutions. Furthermore, we investigate a system of multiple decision-making and explore the impact of learning actors who adjust their trust sensitivities based on solution acceptance and the number of decision-making rounds. Additionally, our approach can be employed to evaluate and refine potential policy modifications. Although we assess potential outcomes using hypothetical actions by individuals, it is essential to emphasize our primary objective of developing a tool that accurately captures real human behavior and fosters improved collaboration in community decision-making. Ultimately, our aim is to enhance the harmony between AI-based recommender systems and human values, promoting a deeper understanding and integration between the two.