MetaClues: Constructivist Learning via Progressive Clues from Generative AI in a Metaverse Environment

This paper introduces MetaClues, an AI-driven framework that integrates generative AI and Metaverse technology to deliver a constructivist learning experience. This innovative system combines image-to-3D generation, large language models, and text-to-image models to create progressive clues within a virtual campus. Learners engage in thematic zones, interact with AI-powered non-playable characters, and manipulate artifact items—creating, fusing, or exchanging them—to steer their own discovery process.At the heart of MetaClues is the goal of uncovering or recreating a secret artifact, a multidisciplinary item rich with educational significance. Learners follow progressive clues tied to a relevance score for each item they possess. This scoring system evaluates an item’s connection to the secret artifact across multiple dimensions, offering insights into its value. To unlock additional clues, learners must create, fuse, or exchange items, refining their understanding of the artifact based on these scores. This adaptive approach strikes a balance between structured, domain-specific learning and open-ended creativity, fostering active engagement.A pilot study with 19 participants showed notable improvements in participation, knowledge retention, and motivation to explore further. Impressively, 84% of novice learners gained a substantive understanding of multidisciplinary concepts. Evaluations using the System Usability Scale (SUS) and the Technology Acceptance Model (TAM) confirmed high usability and educational impact.Using generative AI as a collaborative tool, MetaClues aims to transform learners from passive recipients into creative problem solvers. Our research demonstrated how constructivist principles can be blended with generative AI and immersive technology to enrich multidisciplinary education.