PolicyAI

Democratizing artificial intelligence for improving permitting outcomes and efficiency

Artwork of people walking and riding bikes, surrounded by different forms of energy consumption and creation, covered by an overlay of a computer chip with the acronym AI in the middle.

(Photo composite by Nathan Johnson | Pacific Northwest National Laboratory)

Join our kaggle challenge from now until june 30th

About PolicyAI

Work done under the National Environmental Policy Act (NEPA) has allowed federal agencies to continuously catalog vast amounts of data on federal projects for over 50 years. These data comprise historical and changing environmental resources in various subject matter areas and represent almost every corner of the United States. However, because these data and the documents that house them are not collocated in a single, easily searchable database, much of the information is essentially lost.

Using large language models (LLMs) to inform future reviews by synthesizing information from thousands of existing environmental review documents (e.g., categorical exclusions (CXs), environmental assessments (EAs), and environmental impact statements (EISs)) could result in significant efficiency gains, the identification of otherwise unseen trends, the creation of a comprehensive map, and a catalog of all NEPA efforts.

Moving forward, artificial intelligence (AI) can augment the efforts and expertise of federal agencies to improve and expedite environmental and permitting reviews. For example, AI can be used to access relevant historical environmental review documents and eventually other relevant databases to organize relevant information to facilitate a subject matter expert's review. 

Chart with arrows moving from "Public NEPA Documents" to "Domain Knowledge" to "Domain Understanding" to "Knowledge Discovery"
(Graphic courtesy of Sameera Horawalavithana | Pacific Northwest National Laboratory)

Our Role

This project team was created to serve as the main technical provider in 1) data collection, management, and governance; 2) AI model design, development, and deployment; and 3) safe, secure, and trustworthy technology services to those NEPA subject matter experts and consumers (e.g., government agencies, academia, tribal leaders, and the public) who plan to develop and evaluate the permitting-related tools and applications.

To this end, the team is constructing AI applications to serve as digital research assistants with unprecedented access to past NEPA reviews on various use cases. Some use cases include the following:

  • querying one or multiple environmental review documents for information (e.g., about a location or species) to understand trends over time or to synthesize information for a new licensing action
  • assisting subject matter experts in finding and evaluating information applicable to their review that may be difficult to locate through traditional research methods
  • delineating scoping comments or comments on a draft document (e.g., from the public, other agencies, tribal leaders, and industry) in a consistent manner to assist in efficient and accurate subject matter expert review and response
  • synthesizing draft summaries of comments on issues to provide an inclusive, but brief statement that can be addressed most appropriately by a subject matter expert

These AI tools will help NEPA experts make timely and informed decisions that balance the need for societal progress with environmental protection.
 

Contact Us

Email our team with any questions and/or feedback here. 

Research topics

Artificial Intelligence