GLARE: A Natural Language Interface for Querying Global Explanations

The article presents GLARE, an LLM-based interactive interface designed for querying global explanations of black-box image classifiers using natural language. The system utilizes a core LLM to convert user questions into structured SQL queries, allowing for flexible data aggregation and presenting results as statistics-augmented natural language responses with visualizations. This approach enhances the accessibility and usability of global explanations in explainable AI (XAI), particularly for practitioners seeking targeted insights from complex models.