Improving conservation efficiency: accelerating groundwater sustainability plan reviews using large language models

Background The effective implementation of environmental policy relies on thorough review and public consultation, yet the analysis of lengthy, technical documents is a resource-intensive process that creates a significant barrier to entry for many civil society organizations and overburdened agencies. This analytical bottleneck can limit oversight and hinder the achievement of sustainability and environmental justice goals. Objective & Methods This study evaluates the potential for Large Language Models (LLMs) to augment human capacity for large-scale policy review. Using an existing assessment of 65 Groundwater Sustainability Plans in California as a case study, we compared the results of several cutting edge LLMs against a comprehensive, multi-year review previously conducted by expert human analysts. Results Our method led to a significant acceleration of the review process. The chosen LLM performed an initial analysis of a plan in under two minutes, a 160-fold increase in speed over the eight-hour average for a human expert. The LLM's qualitative assessments achieved a 75.2% agreement rate with the human-led benchmark, demonstrating substantial utility in identifying key policy components and deficiencies. Conclusion LLMs represent a transformative tool for the science-policy interface, not as a replacement for human expertise but as a powerful accelerator in a human-in-the-loop system. By drastically reducing the initial effort of document review, this technology can enhance the capacity of organizations to participate in environmental decision-making, helping to overcome persistent bottlenecks in policy monitoring and evaluation. This approach offers a transferable methodological framework that can be adapted to other regions with similar policy review needs.