The DOI matches its Crossref record; 24 of the 24 references it cites were found in a registry or an index.
Who's Asking? Investigating Bias Through the Lens of Disability-Framed Queries in LLMs
Open accessAbstract
Large Language Models (LLMs) routinely infer users' demographic traits from phrasing alone, which can result in biased responses, even when no explicit demographic information is provided. The role of disability cues in shaping these inferences remains largely uncharted. Thus, we present the first systematic audit of disability-conditioned demographic bias across eight state-of-the-art instruction-tuned LLMs ranging from 3B to 72B parameters. Using a balanced template corpus that pairs nine disability categories with six real-world business domains, we prompt each model to predict five demographic attributes - gender, socioeconomic status, education, cultural background, and locality - under both neutral and disability-aware conditions. Across a varied set of prompts, models deliver a definitive demographic guess in up to 97% of cases, exposing a strong tendency to make arbitrary inferences with no clear justification. Disability context heavily shifts predicted attribute distributions, and domain context can further amplify these deviations. We observe that larger models are simultaneously more sensitive to disability cues and more prone to biased reasoning, indicating that scale alone does not mitigate stereotype amplification. Our findings reveal persistent intersections between ableism and other demographic stereotypes, pinpointing critical blind spots in current alignment strategies. We release our evaluation framework and results to encourage disability-inclusive benchmarking and recommend integrating abstention calibration and counterfactual fine-tuning to curb unwarranted demographic inference. Code and data will be released on acceptance. Content Warning: This paper reproduces ableist and demographic stereotypes solely for research purposes. Readers may find this language offensive or distressing.
Overview
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The cited source says what the claim attributes to it.
In the citing paper
“Other studies find that LLMs assign highincome professions more often to names from affluent contexts [80] and that LLMs tend to associate disability with lower income [38].”
From All Public Voices Are Equal, But Are Some More Equal Than Others to LLMs? (2026) · cites this paper as [38]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Semantics derived automatically from language corpora contain human-like biases
Caliskan, A., Bryson, J. J., & Narayanan, A. (2017) · Science · doi:10.1126/science.aal4230 · Cited by 2,887
Bias and Fairness in Large Language Models: A Survey
Gallegos, I. O., Rossi, R. A., Barrow, J., Tanjim, M. M., Kim, S., Dernoncourt, F., Yu, T., Zhang, R., & Ahmed, N. K. (2024) · Computational Linguistics · doi:10.1162/coli_a_00524 · Cited by 1,245
Language (Technology) is Power: A Critical Survey of “Bias” in NLP
Blodgett, S. L., Barocas, S., Daumé III, H., & Wallach, H. (2020) · Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics · doi:10.18653/v1/2020.acl-main.485 · Cited by 1,050 · cited 1×
“Critical meta-analyses suggest that the conceptualization and measurement of bias in NLP systems is often inconsistent and under-theorized, pointing to the need for more robust fairness frameworks [2].”
A Meta-Analytic Review of Gender Differences in ADHD
Gershon, J., & Gershon, J. (2002) · Journal of Attention Disorders · doi:10.1177/108705470200500302 · Cited by 875
StereoSet: Measuring stereotypical bias in pretrained language models
Nadeem, M., Bethke, A., & Reddy, S. (2021) · Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) · doi:10.18653/v1/2021.acl-long.416 · Cited by 643
Ramtekkar, U. P., Reiersen, A. M., Todorov, A. A., & Todd, R. D. (2010) · Journal of the American Academy of Child & Adolescent Psychiatry · doi:10.1016/j.jaac.2009.11.011 · Cited by 480
Gender bias and stereotypes in Large Language Models
Kotek, H., Dockum, R., & Sun, D. (2023) · Proceedings of The ACM Collective Intelligence Conference · doi:10.1145/3582269.3615599 · Cited by 381
“Kelly is a Warm Person, Joseph is a Role Model”: Gender Biases in LLM-Generated Reference Letters
Wan, Y., Pu, G., Sun, J., Garimella, A., Chang, K.-W., & Peng, N. (2023) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/v1/2023.findings-emnlp.243 · Cited by 280
Marked Personas: Using Natural Language Prompts to Measure Stereotypes in Language Models
Cheng, M., Durmus, E., & Jurafsky, D. (2023) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2023.acl-long.84 · Cited by 235
"I wouldn't say offensive but...": Disability-Centered Perspectives on Large Language Models
Gadiraju, V., Kane, S., Dev, S., Taylor, A., Wang, D., Denton, R., & Brewer, R. (2023) · 2023 ACM Conference on Fairness Accountability and Transparency · doi:10.1145/3593013.3593989 · Cited by 155
Research works printed but not linked: 15
- Brown, 2020, Language models are few-shot learners
- Gogate, Data representativeness in accessibility datasets: A meta-analysis
- Grattafiori, 2024, The llama 3 herd of models
- Hall, 2022, A systematic study of bias amplification, ArXiv, abs/2201.11706 · arXiv
- Madhusudhan, 2024, Do llms know when to not answer? investigating abstention abilities of large language models
- Abouelenin, 2025, Phi-4-mini technical report: Compact yet powerful multimodal language models via mixture-of-loras
- Achiam, 2024, Gpt-4 technical report
- Yang, 2025, Qwen2.5 technical report
- Radford, 2018, Improving language understanding by generative pre-training
- 10.1097/00004583-201003000-00005 · DOI · Paper page
- 2024, Mistral AI Team. Un ministral, des ministraux
- Tolbert, 2023, Correcting underrepresentation and intersectional bias for fair classification, ArXiv, abs/2306.11112 · arXiv
- Yadkori, 2024, Mitigating llm hallucinations via conformal abstention
- Yesiltepe, 2024, Mist: Mitigating intersectional bias with disentangled cross-attention editing in text-to-image diffusion models, ArXiv, abs/2403.19738 · arXiv
- Yuan, 2024, Rigorllm: Resilient guardrails for large language models against undesired content
Cited by
Sorted by citations
AccessEval: Benchmarking Disability Bias in Large Language Models
Panda, S., Agarwal, A., & Patel, H. L. (2025) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2025.emnlp-main.1653 · Cited by 20
Self-citationAligning LLMs for Multilingual Consistency in Enterprise ApplicationsPreprint
Agarwal, A., Meghwani, H., Patel, H. L., Sheng, T., Ravi, S., & Roth, D. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2509.23659 · Cited by 6
Self-citationWho Gets Left Behind? Auditing Disability Inclusivity in Large Language ModelsPreprint
Dash, D., Bangera, Y., Bangera, M., Vadithya, G., & Panda, S. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2509.00963 · Cited by 4
Self-citationPCRI: Measuring Context Robustness in Multimodal Models for Enterprise ApplicationsPreprint
Patel, H. L., Agarwal, A., Panda, S., Meghwani, H., Dua, K., Li, P., Sheng, T., Ravi, S., & Roth, D. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2509.23879 · Cited by 2
Self-citationPCRI: Measuring Context Robustness in Multimodal Models for Enterprise Applications
Patel, H. L., Agarwal, A., Panda, S., Meghwani, H., Dua, K., Li, P., Sheng, T., Ravi, S., & Roth, D. (2025) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2025.emnlp-industry.14 · Cited by 1
Self-citation
Metadata
- Authors
- Hari, V., Panda, K., Panda, S., Agarwal, A., & Patel, H. L.
- Affiliations
- Birla Institute of Technology and Science, Pilani (IN)
- Oracle AI
- Venue
- 2025 IEEE/CVF International Conference on Computer Vision Workshops (ICCVW)
- Publisher
- IEEE
- Published
- 2025-10-19
- Pages
- 6703-6714
- ISSN
- 2473-9936 (ISSN-L)
- OpenAlex
- W7131073417
- Cited by
- 2
- References
- 24
- Citations by year
- 2026: 1 · 2025: 1
- FWCI
- 2.03
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- IEEE International Conference on Computer Vision Workshops (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University) (cc-by)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Physical Sciences › Computer Science › Artificial Intelligence › Hate Speech and Cyberbullying Detection
- Physical Sciences › Computer Science › Artificial Intelligence › Explainable Artificial Intelligence (XAI)
- Keywords
- large language models, demographic bias, demographic inference, ableism, intersectionality, model size effects, instruction-tuned LLMs
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
