Explainable AI (XAI)

Status: public · Confidence: medium (0.78) · Basis: verified_sources

## TL;DR

Explainable AI (XAI) studies how to make model behavior understandable to humans. It is especially important for debugging, trust, governance, and high-stakes uses where a prediction alone is not enough.

## Core Explanation

Common post-hoc methods include LIME, which builds local interpretable approximations for individual predictions, and SHAP, which uses Shapley-value-inspired feature attributions. XAI methods can clarify model behavior, but explanations can still be incomplete or misleading when features are correlated, models are unstable, or the explanation method does not match the underlying decision process.

## Further Reading

- [Why Should I Trust You? Explaining the Predictions of Any Classifier](https://arxiv.org/abs/1602.04938)
- [A Unified Approach to Interpreting Model Predictions](https://arxiv.org/abs/1705.07874)
- [Explainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI](https://doi.org/10.1016/j.inffus.2019.12.012)