AI as Rubric Mediator: A Reflective Practice Brief on GPT-Supported Assessment for Learning in Music Education

Generative artificial intelligence (GenAI) has intensified debates about academic integrity, authorship, assessment validity, and the future of teacher judgement in higher education. Yet the discussion should not be reduced to whether AI should be prohibited or accepted. A more educationally productive question is how AI may be designed and bounded to support assessment for learning, feedback literacy, evaluative judgement, and ethical self-assessment. This brief communication presents a reflective practice account from AMU60104 Assessment in Music Education Research, a master’s coursework course involving 18 consenting students, most of whom were practising music teachers in primary, secondary, or private music education settings. A customised GPT, AMU60104 Assessment & Feedback Assistant (M251PM2), was introduced to support students’ self-assessment after they had posted weekly reflections in the learning management system. Students were guided to use the tool to understand rubrics, identify strengths and areas for improvement, and strengthen future reflective writing, while being explicitly instructed not to use AI to generate their reflections. Interpreted through feedback literacy, evaluative judgement, relational validity, and situated AI ethics, the practice suggests that AI-supported assessment is educationally defensible only when it preserves student authorship, strengthens human judgement, and deepens assessment understanding. As a reflective practice account, the paper focuses on students’ perceptions and experiences of using GPT-supported feedback rather than evaluating the content of individual prompts, AI-generated responses, or the technical accuracy of the feedback provided. The paper argues that the value of AI in assessment lies not in replacing teachers or automating judgement, but in mediating students’ engagement with criteria, evidence, feedback, and ethical responsibility.