Effectiveness of Instruction on the Utilization of Generative AI in Problem Solving: A Case Study of Japanese Junior High School

Abstract

This study investigated the effectiveness of instructional practices involving generative AI (ChatGPT) in supporting problem-solving activities among junior high school students. The instructional design included basic usage of generative AI and its application in real-world problem-solving contexts. The results of measuring changes in students' Chat GPT literacy using the Chat GPT Literacy Scale showed improvements in all subcategories. In addition, analysis of prompt quality and usage patterns revealed that students became more sophisticated in their interactions with the AI. These findings suggest that problem-solving activities using generative AI can contribute to the development of students’ AI literacy. However, the study also identified several challenges, including limited instructional time, individual differences in AI literacy, and the need for more diverse evaluation methods. Moving forward, the study highlights the importance of continuous integration of AI into the curriculum, the use of learning analytics for deeper assessment, and the design of adaptive instructional support tailored to diverse learners.