=Paper=
{{Paper
|id=Vol-3649/Paper9
|storemode=property
|title=Mental Health Stigma across Diverse Genders in Generative Large Language Models - Abstract (abstract)
|pdfUrl=https://ceur-ws.org/Vol-3649/Paper9.pdf
|volume=Vol-3649
|authors=Lucille Njoo,Lee Janzen-Morel,Inna Wanyin Lin,Yulia Tsvetkov
|dblpUrl=https://dblp.org/rec/conf/aaai/NjooJLT24
}}
==Mental Health Stigma across Diverse Genders in Generative Large Language Models - Abstract (abstract)==
<pdf width="1500px">https://ceur-ws.org/Vol-3649/Paper9.pdf</pdf>
<pre>
                                Mental Health Stigma across Diverse Genders in Large
                                Language Models - Abstract
                                Lucille Njoo1,∗,† , Lee Janzen-Morel1,∗,† , Inna Wanyin Lin1 and Yulia Tsvetkov1
                                1
                                    Paul G. Allen School of Computer Science, University of Washington, 185 E Stevens Way NE, Seattle, WA 98195. United States.


                                                  Abstract
                                                  Mental health stigma manifests differently for different genders, often being more associated with women and overlooked
                                                  with men. Prior work in NLP has shown that gendered mental health stigmas are captured in large language models (LLMs).
                                                  However, in the last year, LLMs have changed drastically: newer, generative models not only require different methods for
                                                  measuring bias, but they also have become widely popular in society, interacting with millions of users and increasing the
                                                  stakes of perpetuating gendered mental health stereotypes. In this paper, we examine gendered mental health stigma in
                                                  GPT3.5-Turbo, the model that powers OpenAI’s popular ChatGPT. Building off of prior work, we conduct both quantitative
                                                  and qualitative analyses to measure GPT3.5-Turbo’s bias between binary genders, as well as to explore its behavior around
                                                  non-binary genders, in conversations about mental health. We find that, though GPT3.5-Turbo refrains from explicitly
                                                  assuming gender, it still contains implicit gender biases when asked to complete sentences about mental health, consistently
                                                  preferring female names over male names. Additionally, though GPT3.5-Turbo shows awareness of the nuances of non-binary
                                                  people’s experiences, it often over-fixates on non-binary gender identities in free-response prompts. Our preliminary results
                                                  demonstrate that while modern generative LLMs contain safeguards against blatant gender biases and have progressed
                                                  in their inclusiveness of non-binary identities, they still implicitly encode gendered mental health stigma, and thus risk
                                                  perpetuating harmful stereotypes in mental health contexts.

                                                  Keywords
                                                  NLP, large language models, bias, fairness, gender, mental health, stigma, intersectionality




                                Machine Learning for Cognitive and Mental Health Workshop
                                (ML4CMH), AAAI 2024, Vancouver, BC, Canada
                                ∗
                                     Corresponding authors.
                                †
                                    Authors contributed equally.
                                Envelope-Open lnjoo@cs.washington.edu (L. Njoo); ljanzen@cs.washington.edu
                                (L. Janzen-Morel)
                                            © 2024 Copyright for this paper by its authors. Use permitted under Creative Commons License
                                            Attribution 4.0 International (CC BY 4.0).




CEUR
                  ceur-ws.org
Workshop      ISSN 1613-0073
Proceedings

</pre>