RESEARCHInvestigateNEXT 12 MONTHS
Does Explanation Correctness Matter? Linking Computational XAI Evaluation to Human Understanding
arXiv cs.LG — Machine Learning
Factual evidence
What the source reports
An empirical study of 200 users reveals computational XAI correctness metrics do not reliably predict actual human understanding.
Open sourceOneBench interpretation
Institutional assessment
So what
Flaws in XAI faithfulness metrics challenge standard model validation frameworks used across credit decisioning and risk governance.
Do what
Ask model risk validation teams to evaluate whether human testing is required alongside automated feature attribution metrics.