aiqbench analysis
This research introduces a new benchmark to automatically detect mental health stigma in online text, revealing that current LLMs often fail to distinguish nuanced forms of discrimination from standard sentiment or toxicity. The study demonstrates that without explicit operational rules, large language models tend to overpredict the presence of stigma, highlighting critical gaps in existing safety training data. By releasing annotated datasets and code, this work provides a necessary foundation for improving AI alignment with ethical standards in sensitive social contexts.
Automatic Evaluation of Mental Health Stigma in Online Communication
arXiv:2610.02775v1 Announce Type: new Abstract: Mental health stigma has profoundly harmful impacts but its complexity makes it difficult to evaluate. Stigma may involve explicit derogation, but also subtler forms of blame, fear, paternalistic pity, social distancing, structural exclusion, and discrimination. We introduce a theory-grounded benchmark for automatic evaluation of mental health stigma in online communication, consisting of naturally occurring online news and social media text annotated with a fine-grained taxonomy of stigma across multiple mental health conditions. Our annotation framework comprises a binary stigma-detection task and a multi-level taxonomy covering (i) stigma mode, (ii) domain, and (iii) specific components of certain forms of stigma. We apply this framework to texts mentioning six mental health conditions and evaluate large language models alongside stigma-related classifiers for detecting sentiment, toxicity, and hate speech. Results show that mental health stigma is not well captured by models trained to detect these neighboring constructs, and that LLMs often overpredict stigma unless given explicit operational rules - mirroring the importance of decision rules in human annotation. We release the publicly available part of benchmark, annotations, prototypical exemplars of stigma and code at: https://github.com/jemimakang/mh_stigma.
Abstract:Mental health stigma has profoundly harmful impacts but its complexity makes it difficult to evaluate. Stigma may involve explicit derogation, but also subtler forms of blame, fear, paternalistic pity, social distancing, structural exclusion, and discrimination. We introduce a theory-grounded benchmark for automatic evaluation of mental health stigma in online communication, consisting of naturally occurring online news and social media text annotated with a fine-grained taxonomy of stigma across multiple mental health conditions. Our annotation framework comprises a binary stigma-detection task and a multi-level taxonomy covering (i) stigma mode, (ii) domain, and (iii) specific components of certain forms of stigma. We apply this framework to texts mentioning six mental health conditions and evaluate large language models alongside stigma-related classifiers for detecting sentiment, toxicity, and hate speech. Results show that mental health stigma is not well captured by models trained to detect these neighboring constructs, and that LLMs often overpredict stigma unless given explicit operational rules - mirroring the importance of decision rules in human annotation. We release the publicly available part of benchmark, annotations, prototypical exemplars of stigma and code at: this https URL.
Computation and Language (cs.CL)
arXiv:2610.02775 [cs.CL]
(or arXiv:2610.02775v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2610.02775 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 278 words
Related coverage
Digital States 2026: AI Moves From Experiment to Infrastructure
govtech.com · Sep 24, 2026
LiveLaw Supreme Court Half-Yearly Digest: January - June 2026 [Part 2]
livelaw.in · Sep 13, 2026
The J-Space Debate, Agent Swarms, and Pacing Frontier AI - Digital Minds Newsletter #4
LessWrong · Sep 18, 2026
Models Can't Remember Their Training. Neither Can You.
LessWrong · Jul 19, 2026