TRUST XAI: Model-Agnostic Explanations for AI With a Case Study on IIoT Security

AI systems need explainability to be trusted in high-risk settings: The paper identifies that AI's "black box" nature prevents its adoption in critical applications like industrial infrastructure, medical systems, and financial services—areas where understanding why an AI made a decision is essential for safety and accountability.

TRUST XAI successfully explains AI decisions in industrial security: The proposed method achieved a 98% success rate in explaining AI predictions for cybersecurity threats in industrial IoT systems, demonstrating that effective explanations are achievable even for complex numerical data that typically challenge existing explanation methods.

Model-agnostic explanations work across different AI systems: Unlike explanation tools tied to specific AI models, TRUST can work with any AI system regardless of its underlying architecture, making it more practical for organizations using diverse AI tools across their operations.

Statistical approaches can make AI more transparent: Rather than relying on complex visualization or approximation techniques, TRUST uses statistical analysis to identify which input factors most influence an AI's decisions, offering a mathematically grounded way to understand AI behavior that could be easier for stakeholders to verify and trust.