Abstract:

At the lower secondary level (ages 12–15), art education is expected to foster expressive skills and deepen students' understanding of their aesthetic sensibilities and of gaps between human intention and technologically generated outcomes in art-making in art-making. These gaps are conceptualized in the ART 5.0 and CAPGAP frameworks. Both frameworks emphasize learning through engagement with others. Meanwhile, generative AI has rapidly advanced in recent years, bringing substantial changes to skills and values in art education. While AI has enabled more students to articulate their ideas and worldviews, automated image generation has also evoked feelings of discomfort and confusion.

To address this, we propose ART 5.0 as a framework in which human creativity and technology co-evolve through reflection and dialogue in art education. Each time artistic tools have changed, a new gap has emerged between human intention and generated outcomes. Tools have evolved from hand (ART 1.0), brush (ART 2.0), camera (ART 3.0), and computer (ART 4.0) to generative AI (ART 5.0). In each stage, art education has responded by fostering new skills and values to address these gaps. From this perspective, a starting point for learning in art education lies in the mismatch between intention and outcomes, often experienced as unexpected results or a sense of discomfort. In this study, we develop the CAPGAP model, which organizes gap experiences along three perspectives: Cognitive, Affective, and Participatory. We regard these gaps as learning resources and potential assessment indicators, especially those that emerge as students feel, think, and revise their choices while creating and appreciating with generative AI. Through this lens, we reconsider the nature of art education in ART 5.0. Specifically, we conducted an art education practice with approximately 120 first-year students in a lower secondary school. The lessons were conducted with school and ethics approval. Participation in the analysis was voluntary, and all data were anonymized. The analysis suggested that students reflected across the CAPGAP dimensions. This reflection was prompted by AI-generated images they did not select and by a sense that the results differed slightly from their expectations. Students used these gaps as cues to update their intentions and evaluation criteria, leading them to reselect their expressions. From these findings, this study demonstrates the usefulness of the ART 5.0 and CAPGAP models as frameworks for leveraging gaps between students and technology and for evaluating learning processes in art education. ART 5.0 is expected to provide a foundation for creating new value in art education in the AI era.

Keywords:

Education, Generative AI in education, Learning and assessment.