Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Identifying, Explaining, and Correcting Ableist Language with AI

Kynnedy Simone Smith, Lydia B. Chilton, Danielle Bragg

TL;DR

This two-part study investigates the potential of large language models, specifically ChatGPT, to rectify ableist language and educate users about inclusive communication and highlights the promise and limits of LLMs in handling culturally sensitive content.

Abstract

Ableist language perpetuates harmful stereotypes and exclusion, yet its nuanced nature makes it difficult to recognize and address. Artificial intelligence could serve as a powerful ally in the fight against ableist language, offering tools that detect and suggest alternatives to biased terms. This two-part study investigates the potential of large language models (LLMs), specifically ChatGPT, to rectify ableist language and educate users about inclusive communication. We compared GPT-4o generations with crowdsourced annotations from trained disability community members, then invited disabled participants to evaluate both. Participants reported equal agreement with human and AI annotations but significantly preferred the AI, citing its narrative consistency and accessible style. At the same time, they valued the emotional depth and cultural grounding of human annotations. These findings highlight the promise and limits of LLMs in handling culturally sensitive content. Our contributions include a dataset of nuanced ableism annotations and design considerations for inclusive writing tools.

Identifying, Explaining, and Correcting Ableist Language with AI

TL;DR

This two-part study investigates the potential of large language models, specifically ChatGPT, to rectify ableist language and educate users about inclusive communication and highlights the promise and limits of LLMs in handling culturally sensitive content.

Abstract

Ableist language perpetuates harmful stereotypes and exclusion, yet its nuanced nature makes it difficult to recognize and address. Artificial intelligence could serve as a powerful ally in the fight against ableist language, offering tools that detect and suggest alternatives to biased terms. This two-part study investigates the potential of large language models (LLMs), specifically ChatGPT, to rectify ableist language and educate users about inclusive communication. We compared GPT-4o generations with crowdsourced annotations from trained disability community members, then invited disabled participants to evaluate both. Participants reported equal agreement with human and AI annotations but significantly preferred the AI, citing its narrative consistency and accessible style. At the same time, they valued the emotional depth and cultural grounding of human annotations. These findings highlight the promise and limits of LLMs in handling culturally sensitive content. Our contributions include a dataset of nuanced ableism annotations and design considerations for inclusive writing tools.
Paper Structure (53 sections, 6 figures, 4 tables)

This paper contains 53 sections, 6 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Disability Justice and Foundations of Ableist Language
  4. Ableism in AI Systems and Human–AI Disagreement
  5. Automated Detection and Correction of Biased Language
  6. Study 1 - Collecting Annotations from the Disability Community
  7. Methods
  8. Implications of AI-Generated Stories
  9. Participants
  10. Study Design
  11. Results
  12. Annotation Collection
  13. Inter-Annotator Agreement
  14. Perceived Presence of Ableism
  15. Study 2 - Comparing Human and AI Ableism Annotations
  16. ...and 38 more sections

Figures (6)

  • Figure 1: We use this definition of ableism in our user studies and throughout this paper. It was created by Ashley Eisenmenger, a disability inclusion training specialist at Access Living.
  • Figure 2: The graph on the left describes the percentage of times participants agreed or somewhat agreed with the human or AI identification, explanation, or correction of ableist language across all surveys. The graph on the right shows the percentage of people who preferred the AI ableism annotations, human ableism annotations, or those who had no preference.
  • Figure 3: This table contains keywords and descriptions for the strengths of the human and AI annotators.
  • Figure 4: This table contains keywords and descriptions for the weaknesses of the human and AI annotators.
  • Figure 5: This table contains keywords and descriptions for what participants believed should improve about the ableism annotation system used in this study.
  • ...and 1 more figures