ABSTRACT

This study investigates the potential of generative artificial intelligence (AI), particularly ChatGPT, as a supportive tool for item development and scoring of restricted and extended essay assessments in music education. For this purpose, assessment items appropriate for first-grade middle school music classes were developed using ChatGPT, then implemented and scored in an actual classroom setting. The findings indicate that during the item development phase, the generated items varied in content and scope despite using identical prompts. More advanced versions tended to offer clearer explanations of musical concepts and presented them in a more systematically organized manner. In the scoring phase, using customized GPT improved both efficiency and consistency compared to prompt-based approaches, which required the presentation of multiple materials simultaneously. Notably, incorporating teacher scoring results into the process significantly increased the level of agreement between ChatGPT and teacher scores. These findings highlight both the potential and limitations of generative AI as a tool to support music teachers’ assessment practices. They also suggest that teachers’ critical review and professional judgment are essential for the effective and responsible integration of generative AI in educational assessment.

Keyword :