The Internalized Ableism Inventory: Scale development using a hybrid artificial intelligence and community-based participatory research design.

Objective: This study (a) used OpenAI's GPT-4 large language model to generate an initial item pool for a potential scale measuring internalized ableism, (b) involved disabled community stakeholders in refining the items and prompting additional artificial intelligence-generated items, and (c) psychometrically validated the scale in a large sample of disabled individuals.

Method: Following a series of GPT-4 prompts and iterative community-based participatory research feedback, a tentative item pool of 90 statements was developed. A sample of 409 adults with diverse disabilities completed a survey containing the initial item pool, potentially related scales, and demographic questions.

Results: An exploratory factor analysis helped identify the final 51 items and subscale structure, and a confirmatory factor analysis then provided evidence of excellent factor structure fit. The scale contained eight subscales with Cronbach's αs that ranged from .85 to .97, with an overall total score α of .98. The total score and subscales showed consistent convergent validity with other measures of internalized stigma for chronic illness and anger and frustration with disability.

Conclusion: This study generated for the first time in the known research literature a nuanced, comprehensive, and psychometrically sound scale based on the integration of both artificial intelligence and community-based participatory research methodology: the Internalized Ableism Inventory. The demonstrated methodology generating it has the potential to modernize psychological scale development approaches. (PsycInfo Database Record (c) 2025 APA, all rights reserved).