EmbodiedCreate Teaser

EmbodiedCreate Flowchart


Abstract

Making abstract concepts understandable to children remains challenging. Embodied analogies hold great promise for bridging this gap; however, a single, fixed analogy is often employed in embodied learning environments, which risks surface-level understanding, limiting knowledge transfer for new domains. We present EmbodiedCreate, a novel embodied learning activity that supports analogical learning of AI concepts through the Use-Modify-Create process. It enables students to experience a single embodied analogy, then modify multiple analogies capturing the same concepts, and finally create their own analogies. Through an exploratory study with 17 students (8 high-school, 9 middle-school), we found that: (1) multiple analogies can promote knowledge construction, with learners generating novel mappings between the target and source knowledge; and (2) the creation process supports learners to externalize, critique, and refine their understanding of the AI concepts. Our findings also uncover design implications and challenges of child-AI embodied co-creation for future educational technologies.