Original Broadcast Date: 7/26/2026
Sponsored by Udemy and Carahsoft
Artificial intelligence is creating new opportunities to improve government operations, but every agency must determine where the technology delivers the most value. At the Nuclear Regulatory Commission (NRC), that means identifying AI applications that increase efficiency while maintaining rigorous technical oversight and public confidence.
On Fed Gov Today, Caroline Carusone, Chief AI Officer at the Nuclear Regulatory Commission, discusses how the agency is combining organizational change with AI-enabled tools to modernize its operations. She explains how the NRC reduced its licensing review timeline, why technical experts remain central to every decision, and how the agency is building AI literacy while taking a measured approach to adoption.
Carusone says the NRC's recent modernization efforts extend beyond the implementation of new technology.
The agency recently completed its largest reorganization since its creation and updated its mission statement to adopt what she describes as a more enabling mindset. Those organizational changes help create more direct connections between decision-makers while encouraging employees to approach challenges differently.
At the same time, the agency is identifying high-value, low-risk opportunities to apply AI. "We're really trying to identify where there are high-value but low-risk use cases," Carusone says. Because of the NRC's safety mission, she explains that every implementation must be thoughtful and deliberate.
One example of that approach is the NRC's licensing review process. The agency has reduced the review timeline from approximately four years to 18 months by combining AI-enabled tools with broader operational improvements.
Carusone explains that AI supports the overall composition of safety evaluations and other products, allowing employees to spend less time on repetitive administrative work while focusing more on technical analysis.
She emphasizes, however, that AI does not replace expert judgment. The human is at the beginning, throughout the process, and really bookends the end as well," Carusone says. Subject matter experts remain responsible for the data, decision-making, and final outcomes throughout the review process.
Not every business process requires the same level of change. Carusone says some NRC processes already work well and simply benefit from refinement, while others require a more fundamental redesign.
She points to the agency's oversight and inspection programs as examples in which improved data collection and analysis can help identify areas warranting greater regulatory attention. By becoming more risk-informed, the NRC can focus resources where they have the greatest impact while refining areas that already operate effectively.
As interest in artificial intelligence continues to grow, Carusone says one of the agency's responsibilities is determining whether AI is actually the appropriate solution. She explains that when the NRC collected AI use cases across the organization, many proposals were not truly AI applications. Instead, they represented opportunities to automate repetitive tasks or improve workflows through better data management.
Rather than measuring success by the number of AI deployments, the agency focuses on outcomes. We really measure our AI success around what outcomes we can improve," Carusone says, while also ensuring the agency maintains public confidence.
Alongside technology implementation, the NRC is investing in workforce development.
Carusone says employees across the agency have access to AI tools so they can better understand both the technology's capabilities and its limitations. Training is available for new employees, senior leaders, and the broader workforce, with expectations and metrics established to encourage continued learning.
She believes practical experience is essential, regardless of whether employees are enthusiastic about AI or skeptical of its role. "We have to understand how people are using the AI, and we need to understand...both its promise and its limitations," she says.
Throughout the conversation, Carusone returns to one consistent theme: AI should strengthen expert decision-making rather than replace it. For the NRC, responsible AI adoption means identifying the right problems to solve, applying technology where it provides meaningful value, and ensuring technical experts remain accountable for every decision.
By combining organizational change, measured implementation, and continuous workforce learning, the NRC is building an approach to AI that reflects both its mission and its responsibility to protect public confidence.