In-Progress: Augmenting Explainable AI with LLMs to Enhance User Trust in Intelligent Transportation Systems

Vehicular communication networks increasingly rely on automated systems for misbehavior detection; however, the opacity of machine learning models can diminish user trust. This paper introduces a novel framework that integrates Explainable Artificial Intelligence (XAI) with Large Language Models (LLMs) to deliver transparent, user-friendly explanations for misbehavior detection outcomes. Our approach utilizes XAI to quantify the contribution of individual features, providing clear, visual insights into model decision-making. Additionally, an LLM, augmented with contexts from SHAP and vector database, is employed to convert technical SHAP outputs into accessible natural language explanations. This ensures non-expert users, such as network operators or regulatory personnel, can understand and trust the system's predictions. Experimental evaluations on realistic vehicular network datasets demonstrate that our integrated approach sustains high detection performance while significantly enhancing explainability.