AI use in marking: what schools should know
A practical guide to using AI in marking while keeping teachers accountable for final judgements.
Published 2026-05-19 · Updated 2026-05-19
AI can support marking by drafting feedback, applying rubrics, and flagging responses for review, but teachers should remain accountable for marks used in school decisions.
Use AI as assistance, not final authority
The safest school workflow is teacher-in-the-loop: AI drafts a mark, the teacher checks the evidence, and the school decides how the result may be used.
- Use AI for draft marking, feedback, and workload triage.
- Keep teacher review before reporting marks to pupils, parents, or leaders.
- Avoid describing AI output as an official exam-board result.
Match the risk to the assessment
A low-stakes homework quiz and a mock exam do not carry the same risk. School policy should define where AI-assisted marking is appropriate.
- Class tests and formative checks are the clearest fit.
- Mocks may be suitable when teachers review and moderate carefully.
- High-stakes regulated qualifications require particular caution and official guidance.
Make the workflow explainable
Schools should be able to explain what was uploaded, what the AI drafted, what the teacher changed, and what was exported.
Is AI marking allowed in schools?
Schools can use AI to support some marking workflows, subject to their own policies, data protection duties, and assessment rules. Teacher review is the safest baseline.
Should AI decide final marks?
For MMT workflows, no. The product should be used as AI-assisted draft marking with teacher review before results are used.