Mariangela Mirra

This paper presents the design and pedagogical rationale of an AI-powered avatar developed to mediate Cypriot woven heritage for preschool education. The study is conceptual and design-oriented, focusing on how visual motifs from Cypriot traditional weaving can be reinterpreted through an LLM-based conversational system grounded in retrieval-augmented generation (RAG). The system, named Sophia, transforms textile patterns from the fythiotiko tradition into dialogic narrative prompts that support storytelling, visual exploration, and creative expression in early childhood contexts. The proposed framework is intended to illustrate how artificial intelligence can support the reinterpretation of cultural artifacts as pedagogical resources when embedded within guided-play and museum-based learning approaches. It integrates visual analysis, dialogic storytelling, and artmaking into a structured learning sequence that moves from observation to interpretation and creative transformation. Rather than claiming empirical outcomes, the study positions AI-supported storytelling as a design hypothesis for culturally responsive early childhood education. It argues that AI systems can function as mediational tools between material heritage and children’s imaginative engagement, provided they are developed with pedagogical intentionality, cultural sensitivity, and ethical oversight. The paper concludes by outlining design principles for implementing AI-driven storytelling systems in museums and classroom contexts and identifies the need for future empirical validation in real educational settings.