An AI-Powered Project Approach to Fostering Children’s Computational Thinking: A Quasi-Experimental Study

• In Event: AERA Roundtable Session Sunday 1:45 pm
  In Roundtable Session: AI, Technology, Digital Litercy, and Computational Thinking n Early Childhood

Abstract

This quasi-experimental study evaluated a multimodal GenAI platform ("WonderBuddy") for fostering computational thinking (CT) in 76 kindergarteners (aged 4-5) through project-based learning. Compared to traditional methods, the GenAI approach showed no significant overall advantage. However, moderation analyses revealed stronger benefits for 4-year-olds (Cohen's d = 0.59 vs. 0.03 control) and boys. Qualitative data from teacher interviews identified key facilitators (e.g., multimodal feedback, iterative questioning) and barriers (e.g., technical glitches, workload demands). Findings indicate GenAI can enhance early CT development, particularly for younger children, but effectiveness depends on addressing technical reliability, AI inaccuracy, and providing adequate teacher support. The study offers practical insights for integrating age-appropriate AI tools into early childhood curricula.

Authors

• Weipeng Yang, The Education University of Hong Kong

• Sixuan Xiang, The Education University of Hong Kong