Development of Safe and Secure AI Models for Nuclear Decommissioning Information Management

As nuclear power plants across Europe and around the world reach the end of their operational lives, the complex process of safely and efficiently dismantling these facilities has become a critical task. Decommissioning involves not only technical expertise but also the management of vast amounts of data, ranging from reports and drawings to databases and images. However, this valuable historic and newly created data is often stored in ways that make it difficult to access and utilize effectively for future projects.

This PhD project, hosted by SCK CEN, aims to develop advanced AI models that will revolutionize knowledge transfer in nuclear decommissioning. By leveraging cutting-edge technologies like large language models (LLMs) and multimodal AI, we seek to create tools that make it easier to retrieve valuable information from past or ongoing decommissioning projects and apply these to ongoing or new decommissioning projects. These models will be specifically designed to ensure accuracy, minimize hallucination, and safeguard against potential security threats—critical requirements in the highly regulated and security-conscious nuclear industry.

One of the most innovative elements of this research is the integration of Human-AI collaboration. By combining the strengths of AI with the expertise of human operators, the project aims to create a system that enhances decision-making, security, and efficiency in decommissioning tasks. This work will not only contribute to more effective nuclear decommissioning but also open the door to broader applications of AI in other complex, security-critical fields.

This exciting project offers students the opportunity to work at the intersection of AI and nuclear science, pushing the boundaries of technology while ensuring the safe and secure future of nuclear decommissioning.