Schools Are Using AI to Mark Mock Exams \u2014 Here's What Happened

A Real School, a Real Trial

In April 2026, BBC News reported that Wensleydale School in the Yorkshire Dales had trialled the use of artificial intelligence to mark students' GCSE mock exam papers. The trial covered English and History — subjects with longer written answers where, as headteacher Julia Polley put it, “a lot of subjective marking could happen.”

The school is thought to be one of the only schools in the north of England to have implemented AI marking technology. And their experience offers one of the clearest real-world pictures yet of what happens when a school actually tries it.

This is not a thought experiment or a press release. It is a working school, with real students, real teachers, and real exam papers. Their findings matter — because the questions they set out to answer are the same questions every school, parent, and student is asking.

What the Headteacher Found

Julia Polley was direct about what motivated the trial. The school wanted to test three things: whether AI marking would be accurate, whether it would misinterpret student responses, and what quality of feedback it would produce.

The answer, according to Polley, was positive on all three counts.

Crucially, the school did not hand marking over to AI and walk away. Teachers marked the same papers independently, then compared their results with the AI output. This was a deliberate, controlled comparison — not a shortcut.

Polley was also candid about something that surprised many: in the short term, the trial actually increased teacher workload rather than cutting it, because staff were doing both. The benefit was not immediate time saving — it was confidence in the technology and the quality of feedback students received.

The Bias Question

One of the most striking findings from the Wensleydale trial was about bias. Polley highlighted something that every experienced teacher will recognise: when you know a student, when you know they are trying hard, when you have seen them improve — it is very difficult not to give them the benefit of the doubt.

This is a significant point. Depersonalised marking is a cornerstone of fair assessment — it is why exam boards use anonymous scripts and external examiners. AI marking naturally provides this. It does not know the student, does not know their effort level, and does not adjust its judgement based on context that would not be visible to an external examiner.

For mock exams specifically, where the purpose is to give students an accurate prediction of their real exam performance, unbiased marking is not a nice-to-have — it is the entire point.

The Cost and Practicalities

Wensleydale School set an initial budget of £600 and purchased 1,250 credits at 45p per extended answer. Polley acknowledged this could be seen as expensive, particularly for larger schools with more students.

The trial also required teachers to barcode and individually upload each exam question — a manual process that was “a challenge initially.” This is a common friction point in early AI marking adoption: the technology may be powerful, but the workflow around it needs to be practical for busy teachers.

Moving forward, the school plans to use AI marking for end-of-topic tests and practice exam questions — lower-stakes assessments where the detailed feedback is valuable and the workflow overhead is more manageable.

What the Experts Say

Dr Theocharis Kyriacou, Associate Professor of AI at York St John University, provided important context for where AI marking currently sits in UK education.

Kyriacou identified two clear benefits for students: consistent marking and faster feedback. For teachers, the benefit is time. He noted that teachers work approximately 50+ hours per week and that producing extensive, detailed, personalised feedback is one of the most time-consuming parts of the job.

However, he also warned against completely outsourcing marking to AI. The skill of marking — understanding what students know, where their misconceptions lie, and how to adjust teaching accordingly — is itself a professional competency that teachers should not lose.

He also raised an important perception issue: students and parents may distrust AI marking if they do not understand how it works. Transparency is essential.

The NAHT Position

The National Association of Head Teachers backed the trial. Sarah Hannafin, NAHT Head of Policy, said AI tools “have the potential to deliver opportunities for the education system and reduce administrative burdens on teachers and school leaders — allowing them to focus on areas where their expertise can have the greatest impact.”

Critically, the NAHT position was not unconditional. Hannafin stressed that there must be transparency about when and how AI tools are used, and that they must remain open to challenge. This is the right framework: explore the potential, but do it openly and with proper oversight.

What This Means for Students and Parents

If you are a GCSE student preparing for exams, the Wensleydale trial confirms something important: the quality of AI feedback on essay-based subjects has reached a level where schools are willing to trial it alongside teacher marking. The headteacher described the feedback as detailed and impressive. The staff used it as a sense check against their own marking.

This does not mean AI is replacing your teacher. It means that AI can provide an additional perspective on your work — one that is detailed and does not depend on knowing the student.

For parents, the key takeaway is about transparency and purpose. AI marking is most valuable when it is used as a feedback tool alongside professional teacher judgement — not as a cost-cutting substitute for it. The schools that are getting this right, like Wensleydale, are using both.

Where ReMarkAble AI Fits

We built ReMarkAble AI to address exactly the use case Wensleydale School was exploring: detailed, curriculum-aligned feedback on essay-based GCSE and A-Level work, delivered quickly.

The challenges the school encountered — cost per question, the workflow of uploading individual responses, the need for staff to verify AI output — are challenges we have worked to solve. Students and teachers upload handwritten or typed work directly, receive structured feedback broken down by Assessment Objective, and get an indicative grade aligned to AQA, Edexcel, or OCR mark schemes.

We agree completely with the principle that emerged from this trial: AI marking should be a tool that supports teacher judgement, not one that replaces it. When a student uses ReMarkAble AI to practise essay responses at home, they are getting the kind of detailed, depersonalised feedback that Polley described — and they can bring that feedback to their teacher for further discussion.

The Bigger Picture

The Department for Education published guidelines on AI use in schools in 2025, making it clear that teachers can use AI but must apply professional judgement to check anything it generates. A DfE spokesperson confirmed that “the final responsibility always rests with them and their school or college.”

This is the right regulatory position. AI marking tools are improving rapidly, schools are experimenting, and the results are encouraging. But the adoption curve is still early. As Kyriacou noted, deployment is “patchy and uneven” and mostly limited to low-stakes tasks.

What the Wensleydale trial shows is that when schools approach AI marking thoughtfully — with teacher oversight, clear evaluation criteria, and realistic expectations — the technology delivers. The feedback is detailed, the marking is consistent, and the potential to reduce teacher workload is real.

The question is no longer whether AI can mark essays well enough to be useful. It is how quickly schools, students, and parents will become comfortable using it — and whether the tools available will meet them where they are.

Source: BBC News — “Our teachers use AI to mark mock exams”, published 7 April 2026. Reporting by Emily Johnson.

Frequently Asked Questions