Trustworthy 6G zero-touch network and service management with XAI and LLMs

Achieving Zero-touch Network and Service Management (ZSM) in advanced 6G networks requires intelligent management capable of autonomously detecting and resolving anomalies. However, concerns about trust arise when relying solely on complex Artificial Intelligence (AI) models due to their lack of explainability. In this demo, we present a novel pipeline for trustworthy ZSM leveraging a combination of: (i) AI for anomaly detection using XGBoost; (ii) eXplainable AI (XAI) to provide anomaly root causes with SHAP; and (iii) Large Language Models (LLMs), specifically Llama 2, to generate user-friendly explanations and suggest or apply corrective actions to resolve the anomalies. This approach enhances trust as it facilitates comprehension of ZSM decisions