Slashdot has reported:

Nearly all of the 94 nuclear reactors in the United States have been offered the opportunity to integrate artificial intelligence into their work this year, and most have accepted. The systems are being used to help address documentation and information-management demands, as the industry faces aging equipment and a shrinking workforce.

In August, California-based Atomic Canyon launched NIVA, or the Nuclear Industry Virtual Assistant, developed with nuclear-industry groups. NIVA was pilot-tested at plants operated by Constellation Energy and is now available across the U.S. reactor fleet. The AI-for-nuclear startup Nuclearn says its products are being used by more than 65 U.S. partners, including traditional nuclear plants and NuScale Power Corp., which is working toward advanced small nuclear reactors.

Microsoft and Nvidia have also collaborated on a project to use AI to optimize “the entire life cycle of a nuclear plant, spanning site permitting, design, construction, and continuous operations.” Nuclearn CFO and co-founder Jerrold Vincent described two reasons for the rapid adoption: technology companies investing in AI also want nuclear power revived to supply electricity to data centres, and AI could help with difficult operational challenges in the industry.

Vincent said the intricate regulations and complex management and maintenance systems that make reactors challenging to operate are tasks AI is suited to address. He said the sector is struggling with aging hardware and an aging workforce, and would need machine intelligence to survive and have a meaningful chance at a nuclear revival.

For now, much of the technology’s value lies in handling records. Nuclear plants generate extensive documentation to demonstrate regulatory compliance and record responses to potential problems, ranging from a crack in a sidewalk to a fault in a reactor. Nuclearn’s tools help plant staff search large data repositories more efficiently, file paperwork with the federal government more quickly and find how similar problems were addressed before.

“If you think of it as ChatGPT for nuclear,” Vincent said, “it's actually not a bad place to start.” The workforce pressures add to the need for such tools: the industry has a highly skilled but rapidly shrinking group of workers responsible for keeping older reactors running and potentially bringing new ones online to meet rising electricity demand.

The systems are not directing plant operations. Both the Nuclearn and NIVA large language models used in the nuclear industry report to human users and do not go near power plant operations.