Breakout session (60min) - You Cannot AI Proof Assessments: So What Should You Moderate Instead?

This session draws on my recent professional learning and my own applied practice in designing authentic assessments using generative AI, grounded in the NZQA context. It explores how AI tools can support assessment design, particularly in developing tasks, marking schemes, and rubrics. While AI has significantly improved efficiency, reducing rubric development from hours to minutes, it has also introduced new complexities. Designing assessments that are both fit-for-purpose and robust enough to meet NZQA external moderation expectations remains challenging. The concept of “AI proof” assessment is increasingly unrealistic given the rapid evolution of these tools. A key insight is that AI should be positioned as a support tool, not a substitute for academic judgement or quality assurance. Strong internal processes, particularly pre-and post-assessment moderation, are more critical than ever. This session includes a practical workshop where participants apply a pre-assessment moderation process to a sample assessment. Using a structured template, they will evaluate alignment, evidence sufficiency, and assessor consistency, while identifying tasks that can be easily completed using AI. The session introduces an added moderation lens of AI resiliency, supporting educators to design assessments that require authentic evidence of critical thinking and judgement, even where AI assistance is possible.