Understanding Policy Gaps Affecting Tribal Nations through an AI-Based Platform and Geospatial Analysis

• In Event: Geography, Networks, and Regional Innovation

Abstract

Tribal Nations face disproportionate climate risks while navigating a fragmented policy environment shaped by federal, state, local, and tribal governance. In practice, climate adaptation, disaster resilience, energy, and water-related policies are often dispersed across agencies, jurisdictions, and document types, making them difficult to locate, compare, and use. This creates a practical policy access problem for Tribal communities as well as for planners, policymakers, and researchers seeking to support more equitable adaptation. This project addresses that challenge through an integrated artificial intelligence (AI) tool and a Geographic Information System (GIS) framework focused on the states of Arizona, New Mexico, and Oklahoma.

The study asks three related questions: what barriers to effective climate adaptation policy access are most consistently identified in the literature on Tribal contexts? How are relevant policies distributed across the study region? Can an explainable AI tool improve the retrieval and use of multi-level policy information for Tribal communities and decision-makers?

The project proceeds in two main parts. First, a PRISMA-based literature review identifies recurring barriers in Tribal climate adaptation policy and governance. The review began with 786 records retrieved through database searches covering the period 2008-2025. After filters applied, 270 records remained; 150 articles met the keyword screening threshold, 50 advanced past abstract screening, and 40 studies were included in the final sample. Full-text qualitative coding produced 348 coded quotations, which were synthesized into major themes related to participation, governance, policy structure, and barriers such as limited funding, weak communication, and fragmented institutional capacity. The second phase translates these findings into an applied policy analysis framework. We assembled a structured corpus of 150+ publicly available policy documents from federal, state, local, and Tribal sources across the three states. Documents were standardized by geography, governance level, policy topic, and source type. Spatial analysis in GIS, including hotspot and cluster/outlier methods, was used to identify concentrations and gaps in policy density. In the end, we developed a retrieval-augmented generation AI-based platform / chatbot that supports natural-language policy search, grounded passage retrieval, and traceable links to original documents. The chatbot uses inputs from policy documents and geospatial analysis to help understand available policy at every point in the focus area, as well as provide links and suggestions on how to communicate with it and how relevant those documents could be for a particular query.

Findings suggest that policy density is uneven across the region and does not simply follow the presence of multi-level governance. Although overlapping federal-Tribal governance might be expected to produce denser policy environments, hotspots are concentrated in parts of northern Arizona and New Mexico, as well as northeastern Oklahoma, while southern Arizona and portions of Oklahoma appear as neutral or cold spots. This pattern suggests that metropolitan proximity and institutional concentration may matter more than governance layering alone. The work identifies structural policy access gaps affecting Tribal regions and demonstrates how GIS and explainable AI can support more transparent, place-based, and equity-oriented policy analysis.

Author

• Anton Rozhkov, New York University
  Presenting Author