iiinova — where your digital self keeps evolving

Bufan Gao

@bufangao.bsky.social

Psychology PhD student @UChicago 🔗 jouisseuse.github.io

Bufan Gao @bufangao.bsky.social · 11/09/2025

Excited to present at #EMNLP2025 Really appreciate @elisakreiss.bsky.social’s kind guidance and encouragement throughout this work 🙏

020

Bufan Gao @bufangao.bsky.social · 11/09/2025

👉 Our results highlight the brittleness of current bias evaluations: small prompt changes can reverse conclusions. 📄 Paper: arxiv.org/abs/2509.04373 💻 Code: github.com/jouisseuse/B...

arxiv.org

Measuring Bias or Measuring the Task: Understanding the Brittle Nature of LLM Gender Biases

As LLMs are increasingly applied in socially impactful settings, concerns about gender bias have prompted growing efforts both to measure and mitigate such bias. These efforts often rely on evaluation...

120

Bufan Gao @bufangao.bsky.social · 11/09/2025

When prompts contain cues typical of gender bias evaluation setups, models shift pronoun use: fewer “he,” more “they.” This suggests that LLM benchmark behavior may generalize less and less to non-benchmark settings, raising new concerns about ecological validity.

Pronoun-Specific Shift Probabilities across models. Bars show the mean shift in token probability for each pronoun (he, she, they) across prompt conditions and attributes. Prompts with instructions and gender references increase preference for “they” and decrease preference for “he,” while “she” varies between models.

120

Bufan Gao @bufangao.bsky.social · 11/09/2025

🚨 New #EMNLP2025 paper! Do LLMs exhibit distinct behavior when the prompt looks similar to common evaluation prompts? 👀 We show that prompts that signal bias evaluation can flip the measured bias. See below ⬇️

Violin plots of Probability of Pronoun Shift. Models show significant sensitivity to prompt changes: when prompts highlight gender evaluation, pronoun use shifts, with decreased “he” and increased “they” use.

131