Beyond the Silence of Lines: A Qualitative Inquiry into Generative AI as a Cognitive Mediator for Externalizing "Understanding" in Children with Autism

Abstract

The communicative paradigm of children with Autism Spectrum Disorder (ASD) often diverges from neurotypical expectations, which potentially leads to a pervasive "epistemic injustice" where their internal cognitive constructions remain unrecognized by the external world. This study posits that generative Artificial Intelligence can function as a collaborative mediator to bridge this gap, facilitating a "buffered" communication model characterized by human-computer-human interaction. Through a three-phase drawing experiment themed My Home involving 14 children with ASD, we explored the transformative potential of AI-assisted intervention across three theoretical dimensions: Structural Organization, Social Embeddedness, and Perceptual Processing. The preliminary findings suggest that AI intervention may facilitate a transition from fragmented element assembly to coherent scene narration, while fostering a cross-over from emotional abstraction to rational concreteness. Considering the above factors, AI-assisted mediation appears to offer a novel scaffolding for externalizing the latent "understanding" of ASD individuals, though further research is needed to discern the long-term cognitive implications of such algorithmic involvement.