Effects of integrating AI image recognition and robot game-based learning on computational thinking

This study developed an artificial intelligence (AI) image recognition application integrated with a robot-based game-based learning (GBL) approach to enhance undergraduates’ understanding of computational thinking (CT) and AI concepts. The learning process was guided by experiential learning theory (ELT) or subject-based learning (SBL), combined with interactive gameplay. Sequential behavior pattern analysis and paired sample t-tests were conducted to evaluate the effects of ELT and SBL on learning outcomes, including improvements in learning achievement of CT and AI concepts, computer programming self-efficacy (CPSE), supervised machine learning self-efficacy (MLSE), and reductions in AI anxiety. Analysis of covariance (ANCOVA) and the Johnson-Neyman technique were further applied to compare differences in CPSE and MLSE between the ELT and SBL approaches. The research results show that both learning approaches can increase students’ CT and AI concepts in the robot GBL. The ELT is more suitable for students who initially have lower CT ability and lower self-efficacy. The stage of reflection, abstraction and operation process of ELT in implementing the AI application can enable students to generate discussion, cooperation, and direct manipulation behaviors. Those behavior patterns can enhance their CT and self-efficacy in particular for those who with low CPSE. Conversely, the SBL is more suitable for students who have advanced CT ability and high self-efficacy initially and this study revealed further discussion and suggestions for future studies.