Designing Interactive AI Literacy Systems that Support Children’s Critical and Reflective Engagement with Artificial Intelligence

Abstract

Children are increasingly encountering AI systems in the tools and media that shape how they learn, play, communicate, and create. These technologies offer new opportunities for exploration, creativity, and personalized learning. At the same time, as AI becomes embedded in everyday contexts, there is a growing need for AI literacy that supports children in understanding how these systems operate within broader socio-technical ecosystems. However, many AI systems remain opaque, making it difficult to understand how outputs are generated, what assumptions are embedded in their design, and where their limitations lie. This opacity can hinder the development of accurate mental models and contribute to overestimation of AI capabilities. Supporting children’s AI literacy therefore involves more than explaining how AI works at a technical level. It requires helping children reason about how AI systems produce and represent knowledge, including evaluating AI outputs that may appear credible despite lacking factual grounding, how they reflect and reproduce social values and biases, and how their use depends on material infrastructures with environmental consequences. In this dissertation, I examine how children make sense of AI across epistemic, social, and material dimensions and investigate how interactive systems can support this process. I present the design and study of three systems: AI Puzzlers, CLIP4KIDS, and EcoPrompt. Drawing on classroom observations and participatory design sessions, this dissertation shows how interactive systems can support children’s critical and reflective engagement with AI through iterative experimentation, encounters with AI outputs that surface their own ways of organizing and interpreting information, and connections to their social and cultural worlds.