A mixed-methods study examining how autistic and neurotypical individuals communicate in text-based conversations and how they perceive those interactions. Participants are matched into autistic–autistic, autistic–neurotypical, and neurotypical–neurotypical dyads and complete structured chat sessions followed by annotation of their own conversational moves. The study aims to surface communication asymmetries and design implications for AI-mediated communication support tools.

A survey and computational study exploring what predicts autistic adults' openness to using AI tools for emotion regulation. Findings show that reappraisal ability and difficulties in emotion regulation independently predict AI receptivity, while safety and comprehension concerns act as a separate gatekeeping factor. The work surfaces distinct communication style profiles and examines how AI design preferences vary across them.

Qualitative focus group research exploring how autistic adults envision using AI tools to cope with negative self-talk. Through participatory sessions, we surfaced design visions that center neurodivergent experience — including preferences for non-judgmental, customizable, and transparent AI interactions. The work informs participatory design principles for mental health AI.

A computational social science study of how neurodivergent Reddit communities adopted and adapted ChatGPT — including prompt-sharing practices, community-driven workarounds, and expressed concerns. This work connects community-level AI adoption to accessibility and neurodiversity scholarship.