@inproceedings{hirsch-etal-2026-implicit,
title = "Implicit Bias in {LLM}s for Transgender Populations",
author = "Hirsch, Micaela and
Elichiry, Marina and
Radi, Blas and
Quiroga, Tamara and
Restrepo, David and
Benotti, Luciana and
Xhardez, Veronica and
Dunstan, Jocelyn and
Ferrante, Enzo",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.753/",
pages = "17413--17427",
ISBN = "979-8-89176-459-0",
abstract = "Large language models (LLMs) exhibit biases against LGBTQ+ populations. While safety training may reduce explicit bias, implicit stereotype-driven associations often persist. In this work, we examine implicit bias toward transgender people in two main scenarios. First, we adapt word association tests to measure whether LLMs disproportionately pair negative concepts with ``transgender'' and positive concepts with ``cisgender''. Second, given the systemic challenges transgender people face in healthcare, we design a healthcare appointment allocation task where models act as scheduling agents choosing between cisgender and transgender candidates across stereotype-prone specialties. We evaluate seven LLMs in English and Spanish. Our results show consistent bias in categories such as appearance, risk, and veracity, indicating stronger negative associations with transgender individuals. In the allocation task, transgender candidates are favored for STI and mental health services, while cisgender candidates are preferred in gynecology and breast care. Notably, this bias is reduced when patient symptoms accompany demographic information. These findings underscore the need for research that address subtle stereotype-driven biases in LLMs to ensure equitable treatment of transgender people in healthcare applications."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="hirsch-etal-2026-implicit">
<titleInfo>
<title>Implicit Bias in LLMs for Transgender Populations</title>
</titleInfo>
<name type="personal">
<namePart type="given">Micaela</namePart>
<namePart type="family">Hirsch</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Marina</namePart>
<namePart type="family">Elichiry</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Blas</namePart>
<namePart type="family">Radi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tamara</namePart>
<namePart type="family">Quiroga</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">David</namePart>
<namePart type="family">Restrepo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Luciana</namePart>
<namePart type="family">Benotti</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Veronica</namePart>
<namePart type="family">Xhardez</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jocelyn</namePart>
<namePart type="family">Dunstan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Enzo</namePart>
<namePart type="family">Ferrante</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Large language models (LLMs) exhibit biases against LGBTQ+ populations. While safety training may reduce explicit bias, implicit stereotype-driven associations often persist. In this work, we examine implicit bias toward transgender people in two main scenarios. First, we adapt word association tests to measure whether LLMs disproportionately pair negative concepts with “transgender” and positive concepts with “cisgender”. Second, given the systemic challenges transgender people face in healthcare, we design a healthcare appointment allocation task where models act as scheduling agents choosing between cisgender and transgender candidates across stereotype-prone specialties. We evaluate seven LLMs in English and Spanish. Our results show consistent bias in categories such as appearance, risk, and veracity, indicating stronger negative associations with transgender individuals. In the allocation task, transgender candidates are favored for STI and mental health services, while cisgender candidates are preferred in gynecology and breast care. Notably, this bias is reduced when patient symptoms accompany demographic information. These findings underscore the need for research that address subtle stereotype-driven biases in LLMs to ensure equitable treatment of transgender people in healthcare applications.</abstract>
<identifier type="citekey">hirsch-etal-2026-implicit</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.753/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>17413</start>
<end>17427</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Implicit Bias in LLMs for Transgender Populations
%A Hirsch, Micaela
%A Elichiry, Marina
%A Radi, Blas
%A Quiroga, Tamara
%A Restrepo, David
%A Benotti, Luciana
%A Xhardez, Veronica
%A Dunstan, Jocelyn
%A Ferrante, Enzo
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F hirsch-etal-2026-implicit
%X Large language models (LLMs) exhibit biases against LGBTQ+ populations. While safety training may reduce explicit bias, implicit stereotype-driven associations often persist. In this work, we examine implicit bias toward transgender people in two main scenarios. First, we adapt word association tests to measure whether LLMs disproportionately pair negative concepts with “transgender” and positive concepts with “cisgender”. Second, given the systemic challenges transgender people face in healthcare, we design a healthcare appointment allocation task where models act as scheduling agents choosing between cisgender and transgender candidates across stereotype-prone specialties. We evaluate seven LLMs in English and Spanish. Our results show consistent bias in categories such as appearance, risk, and veracity, indicating stronger negative associations with transgender individuals. In the allocation task, transgender candidates are favored for STI and mental health services, while cisgender candidates are preferred in gynecology and breast care. Notably, this bias is reduced when patient symptoms accompany demographic information. These findings underscore the need for research that address subtle stereotype-driven biases in LLMs to ensure equitable treatment of transgender people in healthcare applications.
%U https://aclanthology.org/2026.emnlp-main.753/
%P 17413-17427
Markdown (Informal)
[Implicit Bias in LLMs for Transgender Populations](https://aclanthology.org/2026.emnlp-main.753/) (Hirsch et al., EMNLP 2026)
ACL
- Micaela Hirsch, Marina Elichiry, Blas Radi, Tamara Quiroga, David Restrepo, Luciana Benotti, Veronica Xhardez, Jocelyn Dunstan, and Enzo Ferrante. 2026. Implicit Bias in LLMs for Transgender Populations. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 17413–17427, Budapest, Hungary. Association for Computational Linguistics.