Academic Journal
Identifying implicit bias in LLM-based chat AI toward people with intellectual disabilities.
| Τίτλος: | Identifying implicit bias in LLM-based chat AI toward people with intellectual disabilities. |
|---|---|
| Συγγραφείς: | Coffey KV; Special Olympics International, USA. Electronic address: kcoffey@specialolympics.org., Krahn GL; Oregon State University, USA. Electronic address: gloria.krahn@oregonstate.edu., Hanley JP; Special Olympics International, USA. Electronic address: jhanley@specialolympics.org., Neely JE; Special Olympics International, USA. Electronic address: jneely@specialolympics.org. |
| Πηγή: | Disability and health journal [Disabil Health J] 2026 Oct; Vol. 19 (4), pp. 102086. Date of Electronic Publication: 2026 May 16. |
| Τύπος έκδοσης: | Journal Article |
| Γλώσσα: | English |
| Στοιχεία περιοδικού: | Publisher: Elsevier Country of Publication: United States NLM ID: 101306633 Publication Model: Print-Electronic Cited Medium: Internet ISSN: 1876-7583 (Electronic) Linking ISSN: 18767583 NLM ISO Abbreviation: Disabil Health J Subsets: MEDLINE |
| Imprint Name(s): | Original Publication: New York : Elsevier |
| Ιατρικοί όροι (MeSH): | Intellectual Disability* , Large Language Models* , Persons with Disabilities* , Artificial Intelligence* , Prejudice*, Humans ; Bias ; Generative Artificial Intelligence ; Adult ; Male ; Female ; Disability Discrimination |
| Περίληψη: | Background: This work investigates the presence of implicit bias in Large Language Model (LLM)-based chat AI models directed toward people with intellectual disabilities (ID). Objective: The study aims to identify and measure representational differences related to people with ID and examine them to identify implicit biases inherent in AI chat generation technologies. Methods: Utilizing the GPT-4-Turbo model, we requested story-generation based on 10 prompt stems with and without descriptors for ID. This process was repeated using four other LLMs (OpenAI GPT-4o, Meta Llama-3-3-70B-Instruct, Anthropic Claude-3-5-Sonnet, and Mistral-Large-2411). The resulting 25,000 computer-generated stories were analyzed using a separate GPT-4-Turbo model instance to detect differences in how people are represented related to themes of bias described in previous literature. Results: Our findings reveal differences in how people are represented between story datasets with and without ID descriptors. These differences go beyond established characteristics of ID and imply the presence of mostly negative implicit biases. Identified differences related to considering people with ID as younger, with themes of paternalism and infantilization; depicting them as more inspirational and symbolic; as needing help more often, being dependent, and being saved; and having a negative perception of them and more hesitation to include them. Conclusions: These implicit biases are considered within the context of past discrimination towards people with ID and highlight the need for diligence against implicit bias towards people with ID in AI development. This research underscores the importance of assessing and mitigating implicit bias in decision-making technologies to prevent future societal harm. (Copyright © 2026 Elsevier Inc. All rights reserved.) |
| Contributed Indexing: | Keywords: Ableism; Artificial intelligence; Implicit bias; Intellectual disabilities; Large language models |
| Entry Date(s): | Date Created: 20260528 Date Completed: 20260904 Latest Revision: 20260904 |
| Update Code: | 20260905 |
| DOI: | 10.1016/j.dhjo.2026.102086 |
| PMID: | 42209384 |
| Βάση Δεδομένων: | MEDLINE |
καταχωρήστε σχόλιο πρώτοι!