Probabilistic Simplified Neutrosophic Sets for Safety Risk Analysis of Complex System Failures under Artificial Intelligence Environment

Abstract

 The safety issues connected with AI-driven operations have increased in importance as 
AI becomes increasingly integrated into complex systems and vital infrastructure. Complex 
systems that depend on automated decision-making and real-time data interpretation, including 
smart grids, autonomous vehicles, and nuclear plant management, can have serious negative 
effects on people, the environment, and the economy if they malfunction. A methodical 
methodology for assessing safety risk elements in intricate systems functioning in AI contexts is 
presented in this paper. System complexity, interpretability, fault diagnostic speed, cybersecurity 
resilience, and vulnerability to ethical risk are among the eight main characteristics that are 
recognized. Ten AI-integrated application scenarios including autonomous cars and medical 
diagnostics are compared using a neutrosophic set methodology. We use the Probabilistic 
Simplified Neutrosophic Sets to overcome uncertainty information. This research suggests 
mitigation techniques unique to AI-specific vulnerabilities and assists in identifying the systems 
most at risk. For designers, operators, and legislators interested in using AI in safety-critical 
systems, the findings offer practical insights

DOI: 10.5281/zenodo.16343820