Rule of Thumb: A New XAI Approach Using Partial Information

A new paper on arXiv proposes 'Rule of Thumb' (RoT) explanations, a novel approach to Explainable Artificial Intelligence (XAI) that identifies the most relevant features for predicting an AI system's behavior for a specific datapoint. The method is designed to enable XAI in zero-shot classification using large language models (LLMs), auditing opaque AI systems without model access, and supporting AI in scientific discovery. RoT meets requirements from leading AI regulations, offers a familiar interface and visualizations for practitioners, is model-agnostic, and is substantially faster than existing alternatives. The paper is categorized under Computer Science > Artificial Intelligence and includes code available at the provided URL. The submission history and references are also included, with the paper available on arXiv with ID 2608.10766.