Decoding AI Complexity: SHAP Textual Explanations via LLM for Improved Model Transparency
Chung-Chian Hsu, I-Zhen Wu, Shih-Mao Liu · 2024
With the continuous advancement of artificial intelligence (AI), particularly in widespread domains such as healthcare and environmental applications, there is an increasing demand for model interpretability. Understanding the decision-making process of models contributes to building trust in them. Hence, the development of Explainable AI (XAI) has become crucial. This study proposes an approach to generate text via a large language model (LLM) for interpretation to enhance the interpretability of SHAP (Shapley Additive exPlanations) plots. The goal is to make the interpretability of model decisions accessible even to non-IT experts through textual explanations.