For the first time, artificial intelligence is being used at a nuclear power plant

AI tool assists Diablo Canyon nuclear plant, enhancing efficiency yet raising oversight debate.

: Diablo Canyon Nuclear Plant in California employs an AI tool developed by Atomic Canyon and powered by NVIDIA H100 chips to help navigate regulatory documents. The AI, named Neutron Enterprise, is designed to improve operational efficiency without making decisions, reducing the time spent searching data by the workforce. Lawmakers express concern about potential expansive AI use and emphasize the need for robust safety measures and regulations. The AI application at the plant marks a pioneering step but is set amidst broader discussions on AI in nuclear facilities under changing regulatory landscapes.

The deployment of artificial intelligence at the Diablo Canyon nuclear power plant in California represents a pioneering application of technology in the nuclear energy sector. The plant, managed by Pacific Gas and Electric (PG&E), has integrated an AI system named Neutron Enterprise to streamline the process of navigating extensive technical reports and regulations from the Nuclear Regulatory Commission. This initiative, developed in partnership with the San Luis Obispo-based AI startup Atomic Canyon, is powered by NVIDIA H100 graphical processors, highlighting the advanced technological infrastructure involved.

PG&E Vice President Maureen Zawalick outlined the phased deployment of the document retrieval AI service, projecting a full operational status by the third quarter of 2025. Neutron Enterprise functions as a data-mining aid, assisting workers in accessing nuclear-specific information more efficiently, thereby reducing the time currently spent navigating databases. Notably, the AI tool is designed to assist rather than make decisions, ensuring that human oversight remains a cornerstone of nuclear plant operations.

The introduction of AI at Diablo Canyon has sparked discussions among lawmakers and AI experts about the necessity for safety measures and regulatory oversight. Tamara Kneese from Data & Society’s Climate, Technology, and Justice program and Democratic Assemblymember Dawn Addis have both emphasized the importance of implementing stringent guardrails to oversee the use of AI in such sensitive environments. This conversation gains complexity in light of regulatory shifts under different political administrations, with former President Trump revoking an AI regulatory executive order from the Biden administration.

Trey Lauderdale, CEO of Atomic Canyon, has expressed commitment to transparency and caution. By leveraging existing data from the Nuclear Regulatory Commission, his company aims to bolster operational efficiency without overstepping regulatory boundaries. However, the potential for broader application of AI technology in nuclear settings remains, albeit with the acknowledgment from Lauderdale and others that regulatory guidance and public oversight are critical.

The developments at Diablo Canyon serve both as a technological milestone and as a catalyst for broader regulatory considerations within the nuclear industry. The plant's current situation is a prototype that could inform future AI applications in nuclear facilities, especially as the industry grapples with the challenge of balancing innovation with safety and regulatory compliance. This case underscores the necessity for ongoing dialogue and development of practical AI standards within nuclear energy contexts.

Sources: CalMatters, The Markup, Southwest Research Institute, Nuclear Regulatory Commission

technologyAIRegulations