Generative artificial intelligence (AI) has created uncertainty in higher education assessment because students can use AI productively for ideation, explanation and technical support, but can also use it to bypass learning. This paper presents the AI Reflection Coefficient Model, an integrity-by-design assessment approach developed within a CDIO-aligned Year 4 Precision Engineering module at Technological University of the Shannon (TUS) Midwest. The model requires students to complete a technical design assignment and submit a structured audio reflection on their use of AI. Reflection quality, assessed using a rubric informed by Gibbs’ Reflective Cycle and the Traffic Light System for AI use, acts as a coefficient influencing the final assignment grade. The study used an exploratory practitioner-inquiry design, combining thematic analysis of student reflections with descriptive analysis of rubric scores. Fifteen students participated in the survey and reflection activity. Findings suggest that structured reflection helped make AI use more visible, encouraged students to recognise AI limitations and supported discussion of ethical boundaries. However, students required further scaffolding to move from procedural description towards deeper ethical reasoning. The paper contributes a practical model for embedding reflective academic integrity practices within AI-supported CDIO assessment.
