Designing a Large Language Model (LLM)-based Chatbot for Neurodivergent Youth

Neurodivergent youth often experience social and communication challenges that can impede their readiness for professional development. While traditional interventions—such as therapist-led coaching and training programs—demonstrate effectiveness, they are frequently constrained by accessibility, scalability, and cost. In this work-in-progress study, we aim to design an AI-driven chatbot by leveraging the latest large language models (LLMs) to support neurodivergent youth in practicing and enhancing their social and communication skills. We plan to (1) design a neurodivergence-informed AI chatbot with therapy-driven prompts and context-sensitive responses by collaborating with domain clinicians and experts, (2) engage neurodivergent youth in the co-design process to ensure the chatbot meet their specific needs, (3) conduct iterative usability testing to assess user engagement, chatbot usability, and user satisfaction, and (4) implement a mixed-methods longitudinal study to evaluate the effectiveness of the chatbot in improving the neurodivergent youth’s social cognition and conversational fluency in their real-life settings. By integrating the latest LLMs into AI-driven chatbot design and using user-centered approaches, this study is expected to contribute to the design of scalable and accessible AI tools that support neurodivergent youth at large.