When Two Marks Diverge: Can AI Serve as a Third Examiner in Master’s Mini-Dissertations?

Abstract Book of the 8th Global Conference on Education and Teaching

Year: 2026

[PDF]

When Two Marks Diverge: Can AI Serve as a Third Examiner in Master’s Mini-Dissertations?

Jacqueline  Wolvaardt, Sean Patrick, Mari van Wyk

ABSTRACT:

Assessment of master’s dissertations often relies on two independent examiners – often an internal examiner and an external examiner. When a substantive discrepancy (greater than 10%) arises between their marks, institutions typically appoint a third examiner, increasing time, cost, and workload, especially in settings where experienced examiners are scarce. This study explores the feasibility of using artificial intelligence (AI) as a third examiner in such cases. The project piloted a structured AI-assisted assessment process aligned to institutional rubrics, assessment criteria, and level descriptors for the mini-dissertations submitted in partial fulfilment of the Master of Public Health at the University of Pretoria (South Africa). AI-generated evaluations were compared with the original examiners’ reports and final moderated outcomes to examine alignment, consistency, transparency, and areas of divergence. Particular attention was paid to the quality of the feedback and any patterns of agreement with either the internal or external examiner. Preliminary findings suggest that AI can provide an impartial, consistent, criterion-referenced benchmark that supports decision-making in cases of examiner disagreement, while not replacing human academic judgment. The paper discusses implications for assessment integrity, examiner calibration, workload reduction, and ethical governance in postgraduate assessment.

Keywords: AI; assessment; ethics; examination; public health