Beyond the Red Pen: AI Efficiently Marks Essays in Minutes for Better Learning Outcomes

A group of universities in the UK – who shall remain nameless – are trialling an Artificial Intelligence tool for essay and test marking. It’s incredible – I’ve seen it working myself.

A lecturer may take 20 minutes to read and mark an essay. Dissertations longer. Their feedback may not be the same as someone else’s so it’s subjective. For A-Levels the sheer volume of handwritten essays teachers have to mark is astounding.

AI can mark them in under a minute, providing scores set by the university to a baseline agreed by teachers or assessors.

Keath.ai – developed by Joey Lin, an engineer and AI expert working towards his PhD at Surrey University – can mark 15000-word essays in three minutes to an accuracy of 85%, providing tailored feedback and working to the same baseline for all the papers it sees. That will soon be 95%, the team says.

When I met Joey, my first question was: “Won’t this anger the university teachers?”

I liked his response.

“Teachers like it because it gives them time to think about what they want the student to work on. But this opens new opportunities for education as a whole.”

“For centuries, we have been applying the same baselines and methods of teaching to everyone who comes through the system. In the same way we are beginning to see tailored medicine for the individual, Artificial Intelligence provides the opportunity for tailored education based on how someone best learns or applies themselves. It means the quality of learning can improve for each person while taking nothing away from the traditional system.”

The Keath team are signing contracts and building steadily in the UK. I am keen to see how the US will see this too.