Agree to Disagree? A Meta-Evaluation of LLM Misgendering
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866909717646802944 |
|---|---|
| author | Subramonian, Arjun Gautam, Vagrant Seshadri, Preethi Klakow, Dietrich Chang, Kai-Wei Sun, Yizhou |
| author_facet | Subramonian, Arjun Gautam, Vagrant Seshadri, Preethi Klakow, Dietrich Chang, Kai-Wei Sun, Yizhou |
| contents | Numerous methods have been proposed to measure LLM misgendering, including probability-based evaluations (e.g., automatically with templatic sentences) and generation-based evaluations (e.g., with automatic heuristics or human validation). However, it has gone unexamined whether these evaluation methods have convergent validity, that is, whether their results align. Therefore, we conduct a systematic meta-evaluation of these methods across three existing datasets for LLM misgendering. We propose a method to transform each dataset to enable parallel probability- and generation-based evaluation. Then, by automatically evaluating a suite of 6 models from 3 families, we find that these methods can disagree with each other at the instance, dataset, and model levels, conflicting on 20.2% of evaluation instances. Finally, with a human evaluation of 2400 LLM generations, we show that misgendering behaviour is complex and goes far beyond pronouns, which automatic evaluations are not currently designed to capture, suggesting essential disagreement with human evaluations. Based on our findings, we provide recommendations for future evaluations of LLM misgendering. Our results are also more widely relevant, as they call into question broader methodological conventions in LLM evaluation, which often assume that different evaluation methods agree. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_17075 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Agree to Disagree? A Meta-Evaluation of LLM Misgendering Subramonian, Arjun Gautam, Vagrant Seshadri, Preethi Klakow, Dietrich Chang, Kai-Wei Sun, Yizhou Computation and Language Computers and Society Numerous methods have been proposed to measure LLM misgendering, including probability-based evaluations (e.g., automatically with templatic sentences) and generation-based evaluations (e.g., with automatic heuristics or human validation). However, it has gone unexamined whether these evaluation methods have convergent validity, that is, whether their results align. Therefore, we conduct a systematic meta-evaluation of these methods across three existing datasets for LLM misgendering. We propose a method to transform each dataset to enable parallel probability- and generation-based evaluation. Then, by automatically evaluating a suite of 6 models from 3 families, we find that these methods can disagree with each other at the instance, dataset, and model levels, conflicting on 20.2% of evaluation instances. Finally, with a human evaluation of 2400 LLM generations, we show that misgendering behaviour is complex and goes far beyond pronouns, which automatic evaluations are not currently designed to capture, suggesting essential disagreement with human evaluations. Based on our findings, we provide recommendations for future evaluations of LLM misgendering. Our results are also more widely relevant, as they call into question broader methodological conventions in LLM evaluation, which often assume that different evaluation methods agree. |
| title | Agree to Disagree? A Meta-Evaluation of LLM Misgendering |
| topic | Computation and Language Computers and Society |
| url | https://arxiv.org/abs/2504.17075 |