Explainable AI for Threat Modelling and Decision Support in Engineering Assets

Abstract

The growing sophistication and interdependency of engineering resources predisposes vulnerable cyber-physical infrastructures
to emerging cyber threats. Explainable Artificial Intelligence (XAI) can be a valuable solution in the field of threat modelling and
the decision-making process because this method will make predictions made on the roots of AI understandable. This paper
addresses the concept of implementing XAI methods to engineer asset threat-assessment systems with a focus on transparency,
accountability, and trust in AI-assisted decisions. The choice of interpretable models, the post-hoc methods of explaining, and
the way AI output can fit in the understanding of the human operators are the main factors to evaluate. The prospects of XAI to
enhance situational awareness, risk prioritization, and proactive response plans are shown in case applications in the energy,
transportation, and industrial systems. The study verifies the existing shortcomings such as model scalability and the trade-off
between interpretability and predictive accuracy and suggests future development directions of auditable and human-centric
AI systems in the protection of critical assets.