Energy Department wants to use artificial intelligence to speed up permitting

With demand for electricity growing thanks to generative AI and related data centers, the Energy Department is testing whether AI can help the government accelerate clean energy production.

The department is spending nearly $20 million to develop and test artificial intelligence-based tools to speed up permitting processes that often take years.

The tools could help government workers leverage historical permitting and environmental data as part of a three-year project called VoltAIc. The hope is that AI will provide better outcomes and speed up decision-making, so clean energy infrastructure can be built faster.

Pacific Northwest National Laboratory, one of the nation’s energy laboratories, has a major project called PolicyAI.

The primary goal is to develop AI-powered software that aims to expand federal oversight under the National Environmental Policy Act, a landmark environmental law that requires agencies to review the significant environmental impacts of major activities, including permitting.

The department is currently testing prototypes internally and has built a data lakehouse that combines elements of a data lake and data warehouse architecture, comprised of historical NEPA documents, said Sai Munikoti, a data scientist at PNNL and the project’s principal investigator.

This lakeside cottage contains information from more than 28,000 Environmental Protection Agency PDF files that align with nearly 3,000 environmental impact reports, one of three potential outcomes of the NEPA process.

These documents detail the potential environmental impacts of a proposed project, as well as potential alternatives. They can be 150 to 300 pages long, according to PNNL.

“We’re trying to add value by making them more accessible,” Munikoti said. The effort includes standardizing the data and adding governance and oversight structures to make the documents more searchable and usable in other AI applications.

Currently, the Lakehouse houses federal environmental impact reports from 2012 to 2023, though the hope is that over time it will be able to add environmental impact assessments and categorical exclusions — other documents that could result from the NEPA review.

Although the EPA keeps all environmental impact reports, there is currently no central database of such documents, said Davie Nguyen, deputy director for state, local and tribal policy in the Office of Energy Policy.

Even “at DOE, we don’t have one place where you can download them all,” Nguyen said. “It’s pretty difficult.”

Adding new documents as they are created would also require collaboration between agencies using different types of technology, but “we’d ​​like to see that functionality eventually,” said Keith Benes, a senior adviser in the Office of Energy Policy.

So far, the team has created a generative, AI-driven semantic search function for the lakehouse that can find relevant documents and summarize search results, Munikoti said. The PNNL team is currently using open-source technology but is working on creating a model tuned for NEPA data.

The search function not only shows users the documents themselves, but also provides a broader overview of a given project, such as related documents, timeline and agencies involved, Munitoki said.

The team is also testing two related pilot programs with energy users while developing agreements with staff from other agencies so they can test the tools as well.

One tool, aimed at helping federal employees navigate projects through NEPA reviews, lets users ask questions and get answers about a single document in a chat interface. The other also uses generative AI to let users “chat” with a group of documents.

The department is using proprietary models such as Gemini and GPT-4 for this purpose, Munikoti said, but the ultimate goal is to move to improved models based on open-source technology for some tasks.

The VoltAIc team is also considering using AI to help state and local reviewers as well, Benes said, potentially through efforts like inventorying all relevant EV charger regulations, for example. Using AI to sort through public comments collected during permitting and environmental reviews is another project the team is considering.

The long-term future of any tools created in the effort—which Energy envisions as an agile sprint to explore possibilities—is not set in stone. But the agency has also “actively started these conversations internally with other agencies about, ‘How could we make this a resource for the future,’” Benes said.

If successful, the work will inform the Biden administration’s efforts to implement groundbreaking pieces of infrastructure and climate legislation, Nguyen said.

“The administration is really focused on, ‘How are we going to make these clean energy investments that we’re making in all the different pieces of legislation right now — how are you going to actually make them happen?’” he said.

He noted that a less complicated and faster permitting environment would also help advance department priorities such as modernizing the U.S. transmission system and electricity infrastructure.

“We need grid infrastructure, clean energy projects that help provide energy security, combat climate change, promote environmental justice, lower energy costs for families… You can’t do all of that if you can’t build the things that help deliver that clean energy,” he said.

The Permitting Council invested $6.1 million in the project as part of a larger set of investments to improve permitting technology across the government. Agencies currently use disparate technologies to power their permitting processes, according to the White House Council on Environmental Quality.

Even though the Energy Department isn’t technically a big agency when it comes to permitting projects — the federal permitting process typically involves multiple agencies, and the details vary from project to project — it’s taking on the problem because of its expertise in AI, Nguyen said.

An AI implementing regulation issued last fall also directed Energy to develop foundation models to streamline permitting and environmental impact assessments.

Part of the work, Benes said, involved talking to potential users of the products.

Broad stakeholder engagement will be key to ensuring the impact of these efforts is truly helpful, said Jessie Mahr, chief technology officer at the nonprofit Environmental Policy Innovation Center.

Efforts to streamline the permitting process should also include removing bureaucratic hurdles that slow down clean energy or recovery efforts in the first place, she said, noting that “permitting is a means to that end.”