In art education, lesson development involves continuous decision-making that is often tacit and difficult to articulate. This study examines how that process can be externalized and restructured through the design and operation of three custom GPTs, each serving a distinct mediating role in the lesson development process. Rather than substituting for teacher judgment, these tools functioned as a space for suspending judgment — enabling teachers to repeatedly pause, verbalize, and reconsider their decisions before committing to them. Through iterative practice, dialogue with the custom GPTs produced structured records of the decision-making process, which in turn supported deeper reflection on lesson design. Notably, this effect extended beyond teachers themselves: the same structures influenced how students verbalized their own decision-making and reviewed their expressive choices, making the learning process visible. These findings suggest that AI tools positioned as mediators — rather than evaluators or generators — hold particular promise for supporting teacher professional development in art education.

View full abstract