How Computer-Based Testing with AI-Assisted Grading Actually Works in EMS

From question bank to a configured test

Before a computer-based test exists, it starts as a question bank — compiled and organised by subject or category, with specific examiners assigned to that bank. A CBT is then configured by drawing from one or more of those banks, giving institutions control over exactly which questions can appear in which exam, and who was responsible for setting them.

Where AI-assisted grading actually helps

Objective, single-answer questions have always been trivial to auto-grade. The harder problem is longer-form responses that need real judgment. EMS's AI-assisted grading is built specifically to reduce the turnaround time on that harder category — surfacing a suggested assessment against a marking scheme — without removing the human examiner from the loop. The examiner reviews, adjusts, and confirms; the system never silently finalises a grade on its own.

Why we drew the line there

Full automation of subjective grading is a genuinely different product with a genuinely different risk profile — one wrong automated call on a student's final grade is a serious problem, not a minor bug. We chose to build EMS as an assistant to examiners rather than a replacement for them, because the accountability for a grade should stay exactly where it already was: with a human examiner who can explain the decision.