AI and Public Policy for Sustainable Natural Resource Governance and Human Health

Abstract

The use of AI in the field of environmental and health administration is outpacing development of the public policy framework to govern responsible AI use. Public administrations employ AI to make decisions in the face of uncertainty and trade-offs. The same AI tools promise to improve water allocation and protection of forests and surveillance of diseases. As dominant technologies promise to create efficiencies, they also become less traceable. These technologies can be used to embed bias and outpace the regulatory systems. This gap in policy is due to rapid AI adoption and tools in public administration. This research adopts a mixed-methods approach that incorporates a comprehensive scan of applications of AI while envisioning and testing a policy-AI framework that integrates environmental health data streams, Internet of Things, and predictive data analytics to support institutional decision-making. A governance-aware feedback mechanism was created to evaluate an ensemble learning hybrid approach that integrates a gradient boosted decision tree for structured data with a health-risk sequence for a streaming health risk. During the ten iterations of the evaluation, a health risk exposure was reduced by 67 percent, while performance improved to 92 percent for the use of resources and prediction of early warnings resulting in improved performance of 94 percent. Results show that the proposed framework outperformed other frameworks in the literature. The indices of governance for transparency, accountability, fairness, and stakeholder participation all demonstrated improvements over the six-year comparative period. It was determined that if responsible AI safeguards are integrated into the technology pipeline as responsible AI, they bring both positive environmental and public health outcomes. This paper concludes that sustainable improvement within resource governance and public health is achieved by considering algorithmic design in parallel with the design of public policy.