September 28, 2025Open Access

How can we trust opaque systems? Criteria for robust explanations in XAI

View Full Paper

Ask AI

Bookmark

Trustworthy explanations must reflect the relevant processes behind deep learning predictions, enhancing user understanding.
Explanatory robustness across different XAI methods ensures consistent and reliable interpretations of model outputs.
The proposed framework formalizes criteria for establishing trust in these opaque systems through explanatory and method robustness.
Significant skepticism exists about current XAI methods, underscoring the need for stronger criteria in evaluation.

Discussion

No takes yet. Share an insight, caveat, or question.

Criteria for explanatory robustness and method robustness in XAI indicate trustworthy predictions from algorithms.

Boge et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark