Design and Application of “AI Art Mentor,” a Generative AI-Based Tool to Support Personalized Feedback in Art Classes
TL;DR
Generative AI–based personalized feedback is conceptualized as a structure responsive to instructional context and students’ current performance and as a process that supports students’ comparison, judgment, and decisions about subsequent artistic expression.
Abstract
This study aimed to design and implement a generative AI–based “AI Art Mentor” to support personalized feedback in art classes and to examine how students interpreted and used AI feedback in relation to their artworks and creative processes. Using Google Gemini’s Gem feature, the AI Art Mentor was configured to align feedback with instructional goals, activity characteristics, students’ current artworks, and stages of production. Data from 90 seventh-grade students, including reflective journals, artworks, AI interaction records, and interviews with five students, were analyzed. The findings showed that students compared AI feedback with their own artworks and judged the appropriateness of suggestions based on their expressive intentions and artwork characteristics. They selectively drew on relevant suggestions or maintained their existing expressions, and some follow-up artworks showed related changes. Students also experienced difficulties translating feedback into material execution and expressed a need for additional support with materials and artistic techniques. This study contributes to art education by conceptualizing generative AI–based personalized feedback as a structure responsive to instructional context and students’ current performance and as a process that supports students’ comparison, judgment, and decisions about subsequent artistic expression.