TY - GEN

T1 - Being Right for Whose Right Reasons?

AU - Thorn Jakobsen, Terne Sasha

AU - Cabello, Laura

AU - Søgaard, Anders

PY - 2023

Y1 - 2023

N2 - Explainability methods are used to benchmark the extent to which model predictions align with human rationales i.e., are ‘right for the right reasons’. Previous work has failed to acknowledge, however, that what counts as a rationale is sometimes subjective. This paper presents what we think is a first of its kind, a collection of human rationale annotations augmented with the annotators demographic information. We cover three datasets spanning sentiment analysis and common-sense reasoning, and six demographic groups (balanced across age and ethnicity). Such data enables us to ask both what demographics our predictions align with and whose reasoning patterns our models’ rationales align with. We find systematic inter-group annotator disagreement and show how 16 Transformer-based models align better with rationales provided by certain demographic groups: We find that models are biased towards aligning best with older and/or white annotators. We zoom in on the effects of model size and model distillation, finding –contrary to our expectations– negative correlations between model size and rationale agreement as well as no evidence that either model size or model distillation improves fairness.

AB - Explainability methods are used to benchmark the extent to which model predictions align with human rationales i.e., are ‘right for the right reasons’. Previous work has failed to acknowledge, however, that what counts as a rationale is sometimes subjective. This paper presents what we think is a first of its kind, a collection of human rationale annotations augmented with the annotators demographic information. We cover three datasets spanning sentiment analysis and common-sense reasoning, and six demographic groups (balanced across age and ethnicity). Such data enables us to ask both what demographics our predictions align with and whose reasoning patterns our models’ rationales align with. We find systematic inter-group annotator disagreement and show how 16 Transformer-based models align better with rationales provided by certain demographic groups: We find that models are biased towards aligning best with older and/or white annotators. We zoom in on the effects of model size and model distillation, finding –contrary to our expectations– negative correlations between model size and rationale agreement as well as no evidence that either model size or model distillation improves fairness.

U2 - 10.18653/v1/2023.acl-long.59

DO - 10.18653/v1/2023.acl-long.59

M3 - Article in proceedings

VL - 1

SP - 1033

EP - 1054

BT - Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics

PB - Association for Computational Linguistics (ACL)

T2 - 61st Annual Meeting of the Association for Computational Linguistics, ACL 2023

Y2 - 9 July 2023 through 14 July 2023

ER -