NeuroBridge: Using Generative AI to Bridge Cross-neurotype Communication Differences through Neurotypical Perspective-taking

Authors: Rukhshan Haroon, Kyle Wigdor, Zihan Yang, Nicole Toumanios, Eileen T. Crehan, and Fahad Dogar

Description: The paper represents an emerging direction in accessibility research that explores how generative AI can foster mutual understanding across different communication styles. NeuroBridge uses large language models to simulate interactions with an autistic communication style and provides feedback to help neurotypical users better understand how language may be interpreted by autistic individuals. Importantly, the work shifts responsibility away from expecting autistic people alone to adapt and instead encourages neurotypical individuals to engage in perspective-taking and reflection about their own communication practices. At the same time, the paper raises important questions about the use of AI to represent disability, noting that users may readily accept AI-generated portrayals as accurate even when they risk oversimplifying the diversity of neurodivergent experiences.