Demographics explain only about 8% of variance in sexism annotations, persona-prompted LLMs do not reliably beat baselines, and SHAP-guided highlighting helps smaller models.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Assessing the Reliability of LLMs Annotations in the Context of Demographic Bias and Model Explanation
Demographics explain only about 8% of variance in sexism annotations, persona-prompted LLMs do not reliably beat baselines, and SHAP-guided highlighting helps smaller models.