Pre-training AI Fundamentals Enhance Game-Based AI Ethics Learning: A Quasi-Experimental Study

Abstract

AI literacy has become crucial for preparing students navigating an AI-mediated world. This requires integrating technical principles with ethical reasoning, yet such comprehensive instructional approaches risk imposing excessive cognitive demands. Pre-training offers a theoretically grounded solution to manage cognitive demands by establishing foundational knowledge before instruction. However, the effects of pre-training in AI literacy education have received limited attention. This quasi-experimental study is on whether video-based pre-training of AI fundamentals enhances the effectiveness of game-based AI ethics instruction compared to game-based learning alone. One hundred fifty-six 7th- and 8th-grade students participated in the study. Results suggested that pre-training AI fundamentals was associated with higher AI ethics capability compared to game-based learning alone (η2ₚ = .028), though classroom-level clustering created statistical uncertainty. No significant difference was observed for expectancy-value beliefs.