AI writing feedback vs AI marking: what's the difference?

What is the difference between AI marking and AI writing feedback?

AI marking applies a mark scheme to finished work and returns a score — a verdict on what the piece was worth. AI writing feedback reads the same work against curriculum criteria and returns strengths and next steps — a suggestion about what the writer does next, reviewed and edited by the teacher before the child sees it.

The two get lumped together because both involve an AI reading a child's writing. But they answer different questions. Marking looks backwards at the piece: how good was it? Feedback looks forwards at the writer: what should they try next? A teacher who has sat writing moderation knows the difference instinctively — it's the difference between a level and a lesson.

What do AI marking tools do?

AI marking tools score finished writing against a mark scheme. Most are built for secondary settings, where mark schemes are published and examiner standardisation materials exist to benchmark against: GCSE English essays, A Level papers, mock exams. Tools in this category — Top Marks AI is a prominent UK example — publish agreement studies against experienced human markers on exam-board materials, and some schools use them to turn a weekend of mock marking into an afternoon.

That is genuinely useful work in the setting it was built for. The mark scheme is the contract: the AI's job is to apply it the way a trained examiner would, and its accuracy can be measured against examiners. The output is a number, because at GCSE the number is the point. Primary writing assessment doesn't work like that — there is no exam script, no examiner, and the child is nine.

What does AI writing feedback do?

AI writing feedback is formative. It reads a piece of writing against the curriculum criteria the teacher chooses — the same ‘pupil can’ statements teachers already assess against — and drafts what a good marker writes in the margin: what this writer did well, with evidence from the page, and the one or two next steps that will move them on.

Three things follow from that design. It surfaces criteria rather than producing a single score, because “working on cohesion between paragraphs” is teachable and a bare number is not. It is teacher-mediated: the output is a draft for the teacher to validate, edit, or override, not a result to be recorded. And it is timed for teaching — feedback that arrives the same lesson can shape the next piece of writing, which is where the learning happens.

In short: AI marking automates a judgement. AI writing feedback drafts a conversation, and the teacher still leads it.

AI marking vs AI writing feedback, side by side

Why do the DfE and the feedback evidence point the formative way?

The evidence base behind feedback is one of the strongest in education. The Education Endowment Foundation's Teaching and Learning Toolkit rates feedback at an average of six months' additional progress, with high evidence security across 155 studies — with the caveat that not all feedback is equal, and that approaches which add workload without giving pupils information they can use should be avoided. Feedback a child can act on is the mechanism; the mark itself moves nothing.

The DfE's guidance on generative AI in education points the same way. It recognises AI can be used for tasks such as feedback, and sets the condition: “Any content produced requires critical judgement to check for appropriateness and accuracy”, with quality remaining “the responsibility of the professional who produced it”. That is a description of teacher-mediated feedback, not of automated verdicts.

And at the statutory end, the STA's KS2 teacher assessment guidance draws a hard line around AI and pupils' independent writing — worth reading in full if you teach Year 6. We've set out the exact wording in AI marking and independent writing.

Where does Howay sit?

Common questions

What is AI writing feedback?

AI writing feedback is formative comment on a piece of writing, generated by AI and reviewed by a teacher before anyone else sees it. Instead of returning a mark, it surfaces the curriculum criteria the writing shows evidence of, names specific strengths, and suggests next steps the writer can act on. The teacher stays in charge: they choose the criteria, edit the output, and decide what reaches the child.

What is the difference between AI marking and AI writing feedback?

AI marking applies a mark scheme to finished work and returns a score or grade — a verdict on what the piece was worth. AI writing feedback reads the same work against curriculum criteria and returns strengths and next steps — a suggestion about what the writer does next, mediated by the teacher. Marking looks backwards at the piece; feedback looks forwards at the writer.

Is AI writing feedback the same as AI grading?

No. AI grading (the usual term in the US) and AI marking both attach a score to finished work, typically against an exam-style mark scheme. AI writing feedback deliberately does not produce a numeric score for the child. It surfaces which curriculum standards the writing evidences and what to work on next, which is the information a primary teacher actually teaches from.

Can AI marking be used in primary schools?

Most AI marking tools are built for secondary settings — GCSE and A Level essays with published mark schemes and examiner standardisation materials. Primary writing is assessed differently: against curriculum 'pupil can' statements, formatively, week by week. That is why primary classrooms tend to need AI writing feedback rather than exam-style AI marking, and why the STA's KS2 guidance puts strict limits on AI anywhere near pupils' independent writing.

Does the DfE allow teachers to use AI for marking and feedback?

The DfE's guidance on generative AI in education recognises that AI can be used for tasks such as feedback, and is clear about the condition: any content produced requires critical judgement to check for appropriateness and accuracy, and the quality of the final output remains the responsibility of the professional who produced it. In other words, AI-assisted feedback is a teacher's tool, not a teacher's replacement.