How to Spot AI in Year 11 Coursework Before Submission

The weeks before a Year 11 coursework deadline are when academic integrity gets tested hardest. Students sit on weeks of half-finished drafts, the deadline panic sets in, and ChatGPT or Claude offers a way out that feels almost reasonable at 11pm. By the time the work hits your inbox, the question is no longer whether to worry about AI — it is how to handle it fairly, before the piece is submitted to the exam board.

This guide explains how to spot AI in Year 11 coursework before submission, what a likelihood score actually tells you, and how to have the conversation with students while there is still time to put things right. The goal is not to catch students out — it is to protect the integrity of their grade and give them a chance to produce work that is genuinely their own.

Why Coursework Deadlines Drive AI Use

Year 11 coursework is high-stakes in a way that classwork is not. The marks count toward a final GCSE grade, the deadline is fixed by an exam board, and missing it carries consequences far heavier than a late homework. That combination — high stakes, hard deadline, weeks of slipping behind — is exactly the pressure point where AI tools become attractive to students who have run out of road.

Most students do not set out to cheat. They start with a legitimate question — "what should I write in the introduction?" — and the boundary between research and substitution slips quietly. By the third or fourth prompt, the AI is doing the writing and the student is doing the copying. Catching this before submission gives both of you a way back. Catching it after the work reaches the exam board does not.

Probabilistic Detection — What a Likelihood Score Really Means

AI detection is probabilistic. No tool — including GradeOrbit's — can give you a verdict that says "100% written by ChatGPT" or "definitely human". What a detection tool produces is a likelihood score: a percentage estimate of how AI-like the writing patterns are, based on sentence structure, word choice variance, and statistical signatures associated with large language model output.

A score of 85% does not mean 85% of the text was written by AI. It means the writing has features statistically consistent with AI-generated text at a high confidence level. The remaining 15% is the room for legitimate cases: a student who naturally writes in a tidy, formal register, a student who has used a spellchecker or grammar tool aggressively, or a student who has revised so heavily that the writing has become "smoothed out".

This is why the score is a starting point for professional judgment, not a verdict. The teacher knows the student. The teacher knows the student's previous writing voice, the kind of vocabulary they use in class, and whether the sudden appearance of "moreover" and "furthermore" is plausible or surprising. The score gives you a reason to look more closely — the looking is still your job.

Reading GradeOrbit's Likelihood Score

GradeOrbit's built-in AI detection tool gives you two model options for any piece of work. The 1-credit model is fast and suitable for a first pass across a class set — useful when you want to triage 30 pieces of coursework and identify which ones warrant a closer look. The 3-credit model is slower and more thorough, designed for the pieces where you want the strongest evidence before having a conversation.

The recommended workflow is to scan the whole class set with the 1-credit model first. Most pieces will come back with low scores and you can move on. The handful that come back high — typically 70% and above — are the ones to re-run on the 3-credit model. If both models return a high likelihood score on the same piece, you have two independent signals pointing the same direction, which is much stronger evidence than one score on its own.

What to Do With a High Score Before Submission

A high likelihood score before submission is a gift. You have time. The exam board has not seen the work. The student's grade is not yet at risk. The right next step is a private conversation — not an accusation, not an immediate referral to the head of department. The aim is to find out what actually happened and give the student a chance to redo the affected sections.

Open the conversation with what you have observed, not with what you have concluded. Something like: "I noticed the writing style in this section is quite different from what you usually produce in class. Can you walk me through how you wrote this paragraph?" gives the student space to explain. If they used AI as a heavy editor or substitute, they will usually say so. If they did not, they will be able to talk you through their drafting process and you have your answer.

If AI use is confirmed, the conversation moves to what comes next. For most schools, the answer is the same: the affected sections are rewritten by the student, in supervised time, before the deadline. The grade is awarded on the rewritten work. The integrity of the qualification is protected and the student learns something more valuable than a borrowed grade — that the shortcut was not actually shorter.

Documenting Your Professional Judgment

Whatever you decide, document it. Save the likelihood scores from both models. Note the date you ran them and the version of the work you analysed. If you had a conversation with the student, write a brief summary of what was said and what was agreed. This is the audit trail that protects you if the case is ever escalated to the exam board, an appeal, or a safeguarding conversation with parents.

For a fuller treatment of the evidence trail and how to talk to students when scores come back high, see our guide to using AI detection as professional evidence and our piece on talking to students about high detection scores.

What GradeOrbit Does Not Store

GradeOrbit never stores student work after analysis. The text you paste in for detection is processed and discarded — it is not retained on our servers, it is not used to train any model, and it does not appear in any other teacher's account. Likelihood scores are returned to you, the work itself is not kept. This matters when the work is a draft of a student's GCSE coursework that you would prefer not to leave sitting on a third-party platform.

Try GradeOrbit's AI Detection This Coursework Season

If your Year 11 coursework deadlines are approaching and you want a way to triage AI use before pieces go to the exam board, GradeOrbit's AI detection tool gives you a likelihood score in seconds and a thorough second-pass option for the pieces that need it. Sign up free at gradeorbit.co.uk — you'll get 10 free credits to try the detection tool and the main marking workflow side by side, with no card required and no auto-renewing trial.