The US Department of Energy (DOE), in collaboration with Idaho National Laboratory (INL), Argonne National Lab (ANL), Microsoft, and Everstar, has demonstrated the use of artificial intelligence (AI) tools to streamline the nuclear regulatory process.
Currently, the nuclear licensing process involves multiple rounds of manual document reviews and minor clerical adjustments, which can take years to complete. The team used AI mapping to convert a safety analysis document required under DOE’s authorisation pathway for advanced reactor demonstrations into US Nuclear Regulatory Commission (NRC) licensing documents for commercial deployment.
DOE said this shows the role AI can play in improving the efficiency and accuracy of nuclear technology licensing, and could one day help to accelerate timelines for the commercial deployment of advanced nuclear reactors.
Everstar’s Gordian AI solution, built on the Microsoft Azure platform, was used to convert the Preliminary Documented Safety Analysis for DOE’s National Reactor Innovation Center’s (NRIC’s) Generic High Temperature Gas Reactor (HTGR) into sections equivalent to an NRC licence application.
The final 208-page document took one day to generate. Typically, the process takes a team of people between four and six weeks to complete the same task. The AI tool also comprehensively identified missing or incomplete information needed to successfully complete an NRC application.
A recent study by NRIC highlighted how AI has the potential to reduce both document development time and regulatory review cycles by as much as 50%, while simultaneously improving accuracy, consistency, and traceability.
AI’s ability to accelerate the development of a licensing application from weeks to days does not mean that nuclear licensing experts are out of a job. The approach maintains a critical principle where experts design, AI accelerates, and experts validate.
Gordian was engineered for nuclear-grade technical work and is equipped with physics and engineering tools, as well as the ability to understand and integrate data through semantic ontology mapping, to ensure that the final output is computed and verified, not inferred. Gordian’s output was subsequently evaluated by an expert for accuracy, missing information, consistency, as well as grammar and structure to ensure that its results were correct and adhered to rigorous professional standards.
DOE said the output “was found to demonstrate quality, rigor, and depth, as well as the tool’s ability to identify and qualify its own gaps in data knowledge”. Kevin Kong, CEO and Founder of Everstar, said: “Nuclear is poised to solve today’s critical energy challenges. We’re excited to partner with INL to meet the moment, working together to accelerate regulatory review and commercialisation.”
Rian Bahran, Deputy Assistant Secretary for Nuclear Reactors noted: “Now is the time to move boldly on AI-accelerated nuclear energy deployment. This partnership… represents more than incremental ‘uplift’ improvements. It has the potential to transform how industry prepares its regulatory submissions and deploys nuclear energy while upholding the highest standards of safety and compliance.”
Carmen Krueger, Corporate Vice President, US Federal at Microsoft said: “Our collaborations with DOE, INL and across the industry are demonstrating how we can effectively bring secure, scalable AI technologies to solve key energy challenges and achieve the broader national and economic security goals envisioned by the Department’s Genesis Mission.”
DOE’s Genesis Mission is a national initiative to unleash a new age of AI-accelerated innovation and discovery. Under the Genesis Mission, DOE recently announced $293m in competitive funding to advance 26 pressing national science and technology challenges, including one focused on expediting nuclear energy deployment.
Looking ahead, the team plans to strengthen and validate their approach. A reviewing agent will evaluate AI-generated documents against NRC guidance to validate that they are ready for submittal. A benchmarking rubric is also being developed to provide a confidence grade for the Gordian’s performance.
The HTGR test case is the latest in a growing list of examples that successfully demonstrates the role AI can play in improving the process. Earlier this year, INL collaborated with Microsoft to deploy a Microsoft Azure AI-based solution to show how advanced AI models can generate engineering and safety analysis reports, a key part of applications for construction permits and operating licenses for nuclear power plants. INL is also developing its own in-house AI tools including potential applications for fuel fabrication facilities.
