Nuclear power plants could soon operate with minimal human staff by using transparent automation systems that make every decision traceable and verifiable, rather than relying on opaque artificial intelligence. This shift addresses a fundamental challenge facing the nuclear industry: small modular reactors (SMRs) deployed in rural and remote locations cannot afford the large operational teams that legacy nuclear plants require, yet safety demands absolute reliability and human oversight.

Why Can't Small Nuclear Reactors Use Traditional Operations Models?

Legacy nuclear power plants operate continuously at full capacity, generating enough revenue to justify maintaining large staffs of skilled operators. But the economics of distributed small modular reactors are fundamentally different. When these plants are scattered across remote locations and operate at variable capacity, the cost of maintaining a large bench of human operators becomes prohibitive. This creates an urgent need for supervised autonomous operations that can handle routine tasks while keeping humans in the loop for critical decisions.

Lauren Fortier, a second-year doctoral student in MIT's Department of Nuclear Science and Engineering, is tackling this problem head-on. Her background gives her unique insight into the challenge. After earning an undergraduate degree in materials science and engineering from Northwestern University, Fortier served as a naval nuclear operator on a U.S. aircraft carrier in the South China Sea, where she supervised nuclear plant operations in one of the most demanding environments imaginable. "It was a unique experience that you don't easily see anywhere else, especially the complete reliance on nuclear power," she explained. "The only way you're moving through the ocean is if you have that nuclear reactor working".

What Makes Fortier's Approach Different From AI-Driven Automation?

During her naval service, Fortier noticed that many plant operations were extremely manually intensive and wondered whether they could be streamlined. When the Navy offered her the opportunity to pursue advanced education, she chose nuclear engineering at MIT to bridge the gap between operational experience and academic research. For her master's degree, she developed a supervisory control system using nuclear plant simulators with realistic thermal hydraulic responses. But the real innovation came when she realized that traditional automation approaches had a critical flaw: they were designed entirely around human operators, leaving little room for machines to contribute effectively.

Fortier's doctoral work focuses on a fundamentally different philosophy. Rather than using machine learning or artificial intelligence, which operate as statistical "black boxes" that are difficult to validate and explain, she is developing automation based on finite state automata. This approach is radically transparent. Every action in the system is event-driven, meaning the automation follows a clear logic: if this condition occurs, then do that action. The system adjusts continuously to current plant conditions and transitions between different operational states with complete clarity.

"We're not using a data-driven statistical approach like machine learning because we do not yet have the tools to validate the operation of such systems," Fortier stated.

Lauren Fortier, PhD Student, Department of Nuclear Science and Engineering, MIT

This distinction matters enormously for nuclear safety. Regulators and operators need to understand exactly why a system made a particular decision. With finite state automata, every decision is traceable and explainable. With machine learning, even the engineers who built the system cannot always explain why it chose a particular action.

How to Design Nuclear Automation Systems That Humans Trust?

• Gradual Autonomy Introduction: Fortier is designing a systematic transition toward autonomy rather than an abrupt shift. When an automated procedure walks operators through steps they would perform anyway, it builds confidence and trust in the system before asking them to rely on it for more complex decisions.
• Human-Machine Collaboration: Instead of replacing human judgment, the system is designed so humans and computers can tag-team, with each doing what it does best. Strategic human intervention is available whenever necessary, and the system clearly indicates when it needs human decision-making.
• Transparent Decision Logic: Every action the automation takes is event-driven and explainable, unlike AI systems that operate as statistical models. This transparency is essential for regulatory approval and operator confidence in safety-critical environments.
• Objective-Oriented Operations: Rather than following predetermined step-by-step procedures, Fortier's control system can determine the sequence of events needed to reach a specific operational objective, adapting to real-time conditions in the plant.

Fortier's research has benefited from collaborations across multiple institutions. Her primary advisor, Sacit Cetiner, holds a joint appointment with MIT and the Idaho National Laboratory (INL), connecting her work directly to real-world nuclear operations. She worked with Katya Le Blanc, a senior human factors scientist at INL, to understand how humans interact with cyber-physical systems and what information operators need when they must take over from an automated system. During a summer internship at Westinghouse in 2025, a leading designer and vendor of current and next-generation nuclear power plants, she tested her autonomous operations concepts against industry standards.

At MIT, Fortier also learned control systems theory from Anuradha Annaswamy, a founder and director of the Active-Adaptive Control Laboratory in the Department of Mechanical Engineering. This expertise helps Fortier translate her operational knowledge into rigorous control systems frameworks. Her other co-advisor, Curtis Smith, the former director of INL's Nuclear Safety and Regulatory Research Division and now KEPCO Professor of the Practice of Nuclear Science and Engineering at MIT, brings decades of regulatory and safety expertise to her work.

The timing of Fortier's research is significant. As nuclear power gains renewed attention as a carbon-free energy source capable of powering data centers and other energy-intensive facilities, the ability to operate small modular reactors safely and economically becomes increasingly important. Her work on transparent, explainable automation offers a path forward that regulators, operators, and the public can understand and trust, without relying on AI systems whose decision-making processes remain opaque.

" }