Stigmatizing Language in Large Language Models for Alcohol and Substance Use Disorders: A Multimodel Evaluation and Prompt Engineering Approach
LLMs frequently generated stigmatizing language when discussing alcohol-related and substance use-related conditions, potentially undermining patient-centered care. However, targeted prompt engineering substantially reduced stigmatizing language occurrences across diverse models. These findings emphasize the need for ongoing model refinement and structured prompting strategies to ensure stigma-free language in health care communication.
Authors
- James Zou (ORCID: https://orcid.org/0000-0001-8880-4764)
- Wei Zhang (ORCID: https://orcid.org/0000-0002-6190-3129)
- Nneka N. Ufere (ORCID: https://orcid.org/0000-0001-8255-9374)
- Christopher Brokus (ORCID: https://orcid.org/0000-0001-5363-1787)
- Yu‐Ting Huang (ORCID: https://orcid.org/0000-0002-4307-5649)
- Kelly Hsu (ORCID: https://orcid.org/0000-0003-2885-6946)
- Yichen Wang (ORCID: https://orcid.org/0000-0003-0723-5913)
- Sarah E. Wakeman (ORCID: https://orcid.org/0000-0002-5475-2713)
Institutions
- Massachusetts General Hospital (US)
Publication Details
- Journal
- Journal of Addiction Medicine
- Published
- 2025-07-24
- DOI
- https://doi.org/10.1097/adm.0000000000001536
- Citations
- 2
- Primary Topic
- Health Policy Implementation Science
- Type
- article
- Field-Weighted Citation Impact
- 5.63