This project explores the integration of Generative Image Models (GIMs) into educational game design to foster social connections between autistic children and their neurotypical peers. The core thesis posits that by shifting from traditional social conditioning methods to creative, collaborative storytelling, digital games can provide a neuro-affirming environment that respects the unique communication styles of autistic individuals. By leveraging the visual-spatial strengths often associated with autism, the project aims to move beyond passive "story-choosing" mechanics toward active "story-creating" experiences that bolster student agency.
The proposed prototype features a two-phase hybrid learning experience where children collaborate to create a digital picture book. Players make visual choices—such as selecting characters, environmental settings, and narrative symbols—which are then rendered into a tangible, printed keepsake. This process is designed to be free from the pressures of verbal communication or rigid social norms, allowing for natural interaction. The methodology relies on qualitative data collection, including on-site behavioral observations, analysis of the co-created digital books, and follow-up questionnaires for caregivers to assess the impact on peer bonding and social engagement.
Currently in the prototype stage, the project utilizes tools like DALL-E, Midjourney, and Adobe Photoshop to generate art assets. While the initial iteration relies on pre-generated choices, future development aims to enable direct, real-time interaction between children and AI to facilitate open-ended storytelling. By providing a structured yet flexible platform, the research seeks to demonstrate that AI-driven creative tools can serve as a bridge for neurodiverse socialization, offering a scalable model for inclusive educational game design that prioritizes creativity over compliance.
