What is No More Marking?
No More Marking is an AI assessment company that helps schools assess pupil writing using AI-enhanced comparative judgement. Our platform combines AI judges, teacher expertise and a statistical model to provide faster assessment, reliable scores and detailed feedback.
Our AI approach builds on years of experience with human comparative judgement. As our explainer video describes, we have assessed more than 3 million pieces of pupil writing using human comparative judgement, with national projects in the UK, USA, Australia and New Zealand.
What is comparative judgement?
Comparative judgement is an assessment method in which judges compare two pieces of work and decide which is better, using a shared assessment question. Many pairwise decisions are combined to estimate the relative quality of each piece of work.
For example, a teacher might read two pieces of pupil writing and choose the one that is better overall. Each piece is compared with several others, so the result draws on a series of decisions rather than a single comparison. Comparative judgement is often abbreviated to CJ; comparative judgment is the US spelling.
How does No More Marking combine AI and teacher judgement?
AI judges compare pairs of writing alongside teacher judges. The statistical model combines their decisions to produce a score for each piece of writing. Schools can choose the proportion of comparisons completed by AI and by people.
Our recommended approach is 90% AI judgements and 10% human judgements. With this approach, every essay is seen twice by humans, and the human decisions provide a check on the AI. Teachers receive a report of disagreements between human and AI judges so they can review the results.
Alongside assessment scores, No More Marking provides personalised pupil feedback reports with specific, pupil-friendly action steps. Teachers receive a separate report with more data and detail to help them plan teaching.
How does comparative judgement work?
Judges see pairs of responses, choose the better response in each pair, and repeat the process. A statistical model combines the decisions into a scale and a rank order.
In No More Marking, both AI and teachers can make these comparisons online using scanned or typed work. Teachers then review the results, the assessment’s reliability and any disagreements between human and AI judges.
How is it different from marking with a rubric?
With a rubric, a marker awards points or levels to an individual response against a set of descriptors. With pairwise comparative judgement, the judge decides which of two responses better meets the assessment question; scores are derived from the combined comparisons.
Judges still need a shared understanding of what they are assessing. For writing, this may mean judging overall quality rather than assigning separate marks for each feature. Teachers can discuss examples of stronger and weaker work to inform their teaching.
Why use comparative judgement instead of AI rubric marking?
When an AI tool scores each pupil’s essay separately against a rubric, the key challenge is consistency across pupils. A rubric describes what good writing looks like, but does not by itself demonstrate that the same standard has been applied to every essay. A convincing explanation of a mark is not evidence that the mark is reliable.
No More Marking builds scores from repeated comparisons between pieces of writing. Judges decide which of two responses is better, each piece is compared with several others, and a statistical model combines the decisions onto a shared scale. Each score draws on multiple judgements about the writing’s quality relative to other work.
The advantage for schools is an assessment they can check. No More Marking reports the reliability of the combined results, and our recommended AI-enhanced workflow includes human judgements alongside AI decisions. Teachers can review disagreements between humans and AI. Our approach combines multiple comparisons, a shared scale and human checks, giving schools evidence to evaluate the results before using them.
What can comparative judgement assess?
Comparative judgement can be used for open-ended work where quality requires professional judgement, such as essays and other extended responses. No More Marking focuses on helping schools assess writing.
The assessment question should make clear what judges are comparing. A question such as ‘Which is the better piece of writing?’ asks about overall writing quality. A different question would define a different assessment.
Is comparative judgement reliable?
Reliability describes how stable assessment scores are likely to be if more judging is done or different judges take part. No More Marking reports reliability on a scale from 0 to 1, with higher values indicating more stable scores.
Reliability depends on the amount of judging and agreement between judges. No More Marking typically aims for a value above 0.8, but the precision needed depends on how the results will be used.
How much marking time can AI-enhanced comparative judgement save?
Our explainer video gives an example of 30 essays that would take two hours to mark with a rubric. In that example, human-only comparative judgement takes one hour. With AI completing 90% of the judgements, the teacher’s judging time falls to six minutes: a 95% reduction compared with the original two hours.
The six-minute figure is an illustration of teacher judging time, based on the video’s assumptions. The time needed depends on the length of the writing, how quickly teachers make comparisons and how much feedback they add. Our helpdesk gives a longer worked example using different assumptions.