Abstract
CritiQAL is a structured viva-based assessment design built using Generative Artificial Intelligence (GenAI) to support assurance of learning. The key idea behind CritiQAL is to exercise students’ evaluative judgement – that is, the ability to assess and evaluate the quality of explanations or solutions. It was implemented in Trimester 1, 2026 in SIT796 Reinforcement Learning, a technical unit in Artificial Intelligence at Deakin University with a cohort of 139 postgraduate students. The tool was trained on unit materials to generate questions and suggest answers aligned to unit content, with parts of a suggested answer intentionally constructed to be incorrect. Students were tasked with answering questions and critiquing suggested answers, shifting the task from simple recall towards evaluation, justification and explanation. CritiQAL was used to conduct a viva task from week six onwards, with a full, half or zero-point rubric, where students were required to accumulate five points by Week 11. This paper outlines the rationale, design and implementation of CritiQAL as a practical response for assurance of learning in a context increasingly shaped by GenAI.
Keywords
Assurance of learning, evaluative judgement, oral assessment, viva, Generative AI.
