Adapting Human Reliability Analysis to Autonomous Control for Small Modular Reactors

The increasing interest in autonomous and AI-supported control systems for Small Modular Reactors (SMRs) introduces profound changes to the future role, responsibilities, and actions of human operators. While higher levels of automation promise enhanced operational efficiency and reduced staffing, they also reshape traditional human-system interaction patterns, creating new pathways for human error and challenging the applicability of existing Human Reliability Analysis (HRA) approaches. Current HRA methods-developed for operator-centric control architectures-struggle to capture the dynamic, supervisory, and intervention-based roles that humans assume in highly automated, AI-assisted environments. This paper examines the emerging human reliability challenges associated with AI-enabled autonomous control in SMRs, focusing on key aspects as shifts in situational awareness, trust calibration, and transitions of control between automated and human agents. We review the suitability of established HRA methodologies, identify gaps in modelling cognitive load and humanautomation interaction, and propose updates and methodological adaptations needed to evaluate human performance within this new operational paradigm. The aim is to provide a structured foundation for future HRA approaches capable of supporting licensing, safety analysis, and design of AI-driven control systems in next generation nuclear reactors.