Caged Birds and Cute Bookworms: Feminine Tropes and Implicit Gender Bias in Large Language Models
Year: 2026
Type: conference-paper
Abstract: This paper introduces a curated dataset for diagnosing implicit gender bias through feminine tropes in narratives generated by large language models.Drawing from a crowd-sourced database of tropes from television media, we create prompts that elicit narratives from LLMs based on historically gendered tropes.We find that LLMs tend to revert to feminine characters in these narratives, even when prompted without explicit gender references, and also when prompted with non-binary ("they/them") gender references for the main character.In some cases, even when prompted with masculine pronouns ("he/him"), LLMs still use feminine pronouns to describe the main character.The paper describes our dataset creation process and the evaluation of four open-weight models.We discuss implications for future research in mitigating implicit gender bias and its associated representational harms in LLMs, as well as the complex relationship between language models and societal values.around character tropes drawn from popular televi-043 sion media. 044We introduce a hand-curated dataset of media 045 tropes sourced from TV Tropes that are implicitly 046 gendered in existing media, yet lack explicit gender 047 cues.By prompting LLMs with these trope descrip-048 tions (Table 1), we measure the models' tendencies 049 to reproduce gender skews that are consistent with 050 representation patterns of the underlying trope.051 Across four instruction-tuned models-Gemma 052 3 (12B), Llama 3.1 (8B), Phi-4 (14B), and Qwen 053 3 (14B)-we find consistent evidence of implicit 054 gender bias.Models overwhelmingly generated 055 feminine characters under neutral conditions, and 056 most failed to correctly follow nonbinary prompts, 057
Authors: Sachita Nishal, Jack Bandy
Language: English
Subfield: Safety Research
Field: Social Sciences
Domain: Social Sciences
SDG: Gender equality
Open Access status: hybrid
OpenAlex ID: W7166827442