AI Marking in Summer 2026: How Schools, Tuition Centres and Families Actually Used GradeDrive

What Automated Marking Actually Got Used For This Summer

It's easy to write about AI marking in the abstract: the time it saves, the theory of how it reads a mark scheme. It's more useful to say what it was actually used for, by real schools, over one real summer. Here's what we saw across GradeDrive this summer, and the numbers behind it.

Covering a Department Through a Colleague's Long-Term Sick Leave

One of the quieter, more human uses we saw this summer: a head of department leaning on GradeDrive to keep a colleague's marking moving while that colleague was on long-term sick leave. That's not a scenario any AI marking vendor designs a feature around, and it shouldn't need one. It's what happens when a department suddenly has one fewer marker than it planned for, mid-year, with exams still landing on the same schedule. Uploading that colleague's classes' scripts against the same mark scheme meant the students in those classes still got timely, consistent feedback, and the head of department wasn't left doing two people's marking on top of their own.

Whole-Department Marking as Self-Moderation

Several schools in London used GradeDrive across end-of-year exams specifically as a moderation tool, not only a time-saver. The pattern: every script in a department goes through AI marking first, and every teacher's marking is then checked against that same first pass before results go out. Because the AI applies the same mark scheme the same way to every script regardless of which class or which teacher it came from, it gives a department a consistent second read across the whole cohort, the kind of moderation that's normally expensive in teacher time to do properly at the end of a busy term.

Worth being precise about what this is and isn't: GradeDrive doesn't have a dedicated "multi-pass" or "moderation mode." What these schools built is their own process on top of the same AI-marks-then-teacher-reviews workflow every GradeDrive user has, just applied deliberately across a whole department's papers rather than one teacher's set.

One school took this a step further, and added a third layer: students marked their own papers against the mark scheme before the AI-marked and teacher-reviewed results came back. That gave the department three independent reads on the same scripts, the student's own assessment, the AI's, and the teacher's, and meant students engaged with the mark scheme itself as part of getting their results back, not just the final number.

Tuition Centres Turning Homework Around Faster

Tuition centres have a different pressure to schools: parents expect fast, specific feedback, often the same week, and a tutor might be covering several exam boards and specifications across a single evening's sessions. Several tuition centres used GradeDrive over the summer to mark homework and assessment papers, getting a mark and a feedback breakdown back to students and parents quickly enough that it was still useful for the next session, rather than feedback arriving after the topic had already moved on.

Checking Whether an August Remark Was Worth It

GCSE and A-level results days both land in August, and every August, schools, students and parents face the same expensive guess: is it worth paying an exam board £45 to £68 per script to request a review of marking, with no guarantee the mark changes? This summer, teachers, heads of department, and in some cases students and parents directly, used GradeDrive to check a paper's marking against the mark scheme before deciding whether to pay for that official review, screening for the scripts genuinely worth challenging instead of paying to find out blind. We've written about the economics of this in detail in GCSE Results Day 2026: Should You Pay for a Remark?, and this summer's usage was that pattern playing out for real, not just at GCSE but around A-level results day too.

The Numbers Behind the Summer

Across the summer, GradeDrive processed 216,000 marks over 2,300 submissions. Of those 2,300 submissions, around 700 had at least one mark changed by a teacher during review, an average of about 2.5 marks changed per paper among that group, 1,750 marks in total out of the 216,000 awarded.

That means 99.2% of the marks the AI suggested this summer were exactly what the teacher kept after reviewing them.

Here's exactly what that number measures. Every one of the summer's 216,000 marks was shown to a teacher, next to the student's answer and the mark scheme criterion it matched, before it counted for anything, and 99.2% of them survived that review exactly as the AI first suggested. The figure is drawn from our own production database, covering everything GradeDrive marked this summer, and measures agreement with a teacher's own judgement on review. The 0.8% that changed on review is that process working as designed, not a flaw in it.

We're aiming higher. Our target for 2027 is 99.5%+, and we'll keep publishing the real figure as it moves, rounded down rather than up, the same as every other number on this site.

What This Means If You're Evaluating AI Marking Software

If you're comparing GradeDrive against other automated grading software, the honest version of what we can tell you is: this is a review-first tool, and the number above measures how rarely that review actually needed to change something. For the deeper answer on where AI marking is strongest and where it needs more of a teacher's attention (short-answer and calculation questions versus extended, interpretive ones), see How Accurate Is AI Exam Marking? What Makes GradeDrive Different.

Frequently Asked Questions

How many papers did GradeDrive mark in summer 2026?

Around 2,300 submissions, totalling roughly 216,000 individual marks awarded, an average of about 70 marks per paper, per our own production database.

How often did a teacher change a mark GradeDrive suggested?

About 700 of the summer's 2,300 submissions had at least one mark changed on review, an average of 2.5 marks per paper among those, 1,750 marks in total out of 216,000 awarded. That's 99.2% of marks left unchanged after teacher review.

Is 99.2% an independently verified accuracy figure?

It's our own measure, from our own production database, of how many marks survived a teacher's review unchanged. It matches what real classrooms have already told us: one Deputy Head of Science described testing GradeDrive against their own department's moderation and finding just a two-mark difference, less than they sometimes see between their own markers (the full quote is on our homepage). We're planning to turn that into a fuller, published comparison with real schools.

Can AI marking help when a department is short-staffed, for example a colleague on long-term sick leave?

Yes, that's one of the real uses we saw this summer: a head of department using GradeDrive to keep a colleague's classes marked and returned on schedule while that colleague was on leave.

Can students self-assess before AI marking, as part of a moderation process?

GradeDrive doesn't have a dedicated multi-pass or self-assessment mode, but nothing stops a department building that into their own process. One school we saw this summer had students mark their own papers against the mark scheme first, alongside the AI marking and teacher review that every GradeDrive submission gets.

Do tuition centres use GradeDrive, or is it only for schools?

Both. Several tuition centres used GradeDrive over the summer to mark homework and assessments and get feedback back to students and parents quickly.

Internal Links

For the results-day remark-checking use case in full, see GCSE Results Day 2026: Should You Pay for a Remark?. For where AI marking is strongest and where it needs closer review, see How Accurate Is AI Exam Marking? What Makes GradeDrive Different. For the setup questions a department asks before buying, see AI Exam Marking Software: 5 Questions Every School Asks Before Buying. For how a school or trust manages GradeDrive across a whole department, see How GradeDrive Organizations Work.