Faithfulness Without Anchors: A Position Paper on the Structural Misalignments of Explainable AI Faithfulness Metrics for Feature Importance
Abstract
Explainable Artificial Intelligence (XAI) tools have often been presented as methods to increase trust in machine learning models by making aspects of their algorithms more human-understandable. However, evidence accumulated over the last decade points to an overall failure of this objective, instead promoting overreliance on incorrect predictions, which can subsequently lead to algorithmic aversion. Rather, XAI appears to function primarily as a debugging tool, useful for obtaining anecdotal insights into, for example, features learned by the model. One of the main issues underlying this limitation is that XAI methods often oversimplify the inner workings of the model, resulting in misalignment between the explanation and the model itself - a behavior referred to as unfaithfulness or infidelity. Faithfulness can be interpreted as the degree of approximation between an explanation and the model. Computing faithfulness in a rigorous manner would thus make it possible to quantify one of the various elements of trustworthiness of an explanation. However, existing metrics present significant limitations, making this quantification either infeasible or computationally expensive. In this position paper, we outline five major shortcomings of current faithfulness metrics and propose four possible avenues for research to bolster their reliability.
Related
- Ab-faithfulness of absolute Galois groups in number theory
- A Hybrid Structural Modelling and Explainable Machine Learning Framework for Human–AI Strategic Decision Support Systems
- Operationalising Weaker Faithfulness: Recovery, Ambiguity and Computational Limits
- Monadic Second-Order Logic in HOL: Deep and Shallow with Automated Faithfulness (Extended Preprint)
- Just add noise: Debiasing tree-based variable importance in mixed data
BibTeX
@misc{zullich2026faithfulness,
title = {{Faithfulness Without Anchors: A Position Paper on the Structural Misalignments of Explainable AI Faithfulness Metrics for Feature Importance}},
author = {Marco Zullich and Emily Schiller and Ivan Gentile and David Dembinsky and Adriano Lucieri and Kilian Göller and Steffen Seitz},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.12493},
eprint = {2610.12493},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.12493},
note = {CC BY-NC-SA 4.0},
}
RIS
TY - UNPB
TI - Faithfulness Without Anchors: A Position Paper on the Structural Misalignments of Explainable AI Faithfulness Metrics for Feature Importance
AU - Marco Zullich
AU - Emily Schiller
AU - Ivan Gentile
AU - David Dembinsky
AU - Adriano Lucieri
AU - Kilian Göller
AU - Steffen Seitz
PY - 2026
DA - 2026/10/01
PB - arXiv
DO - 10.48550/arxiv.2610.12493
UR - https://arxiv.org/abs/2610.12493
AB - Explainable Artificial Intelligence (XAI) tools have often been presented as methods to increase trust in machine learning models by making aspects of their algorithms more human-understandable. However, evidence accumulated over the last decade points to an overall failure of this objective, instead promoting overreliance on incorrect predictions, which can subsequently lead to algorithmic aversion. Rather, XAI appears to function primarily as a debugging tool, useful for obtaining anecdotal insights into, for example, features learned by the model. One of the main issues underlying this limitation is that XAI methods often oversimplify the inner workings of the model, resulting in misalignment between the explanation and the model itself - a behavior referred to as unfaithfulness or infidelity. Faithfulness can be interpreted as the degree of approximation between an explanation and the model. Computing faithfulness in a rigorous manner would thus make it possible to quantify one of the various elements of trustworthiness of an explanation. However, existing metrics present significant limitations, making this quantification either infeasible or computationally expensive. In this position paper, we outline five major shortcomings of current faithfulness metrics and propose four possible avenues for research to bolster their reliability.
ER -