The Disagreement Problem in Faithfulness Metrics

TL;DR

This paper investigates the inconsistency among existing faithfulness metrics for local explanations in XAI, highlighting the challenge for users in selecting the most reliable explanation methods due to metric disagreement.

Contribution

It provides a comparative analysis of faithfulness metrics, revealing their disagreements and emphasizing the need for clearer evaluation standards in XAI explanations.

Findings

Current faithfulness metrics do not agree on explanations' quality.

Users face difficulty in selecting faithful explanations due to metric discrepancies.

The study underscores the need for standardized evaluation in XAI.

Abstract

The field of explainable artificial intelligence (XAI) aims to explain how black-box machine learning models work. Much of the work centers around the holy grail of providing post-hoc feature attributions to any model architecture. While the pace of innovation around novel methods has slowed down, the question remains of how to choose a method, and how to make it fit for purpose. Recently, efforts around benchmarking XAI methods have suggested metrics for that purpose -- but there are many choices. That bounty of choice still leaves an end user unclear on how to proceed. This paper focuses on comparing metrics with the aim of measuring faithfulness of local explanations on tabular classification problems -- and shows that the current metrics don't agree; leaving users unsure how to choose the most faithful explanations.