Illustrating Inclusion with Artificial Intelligence. Critical Issues and Potential of Generative Prompts in Educational Contexts

Abstract

Images are the first things that children learn to “read” from an early age, through a very early decoding process, which little by little becomes automatic. For children one of the main sources of images is represented by the encounter with fairy tales and illustrated stories, in which the figures comment on and integrate the text (when, indeed, they do not replace it). Based on these premises, this article aims to contribute to the emerging discourse on the intersection between artificial intelligence (A.I.) and education, particularly in the context of children’s literature, since storytelling has always been a cornerstone of education, given its fundamental role in shaping the way individuals understand the world around them and develop social norms accordingly. If we are wondering whether generative artificial intelligence tools, such as ChatGPT, can currently understand and interpret fairy tales well enough to autonomously produce illustrations that capture the essence of these stories, the answer is no. In this contribution we will analyze the reason why, with related educational implications, taking inspiration from the limits and prejudices of visual representation of disability in the AI Text-to-Image (T2I) results, emerged during the Special Education-Children’s Services laboratory of Niccolò Cusano University dedicated to the creation of illustrated fairy tales.