AI for Business
Marking Rubric & Assessment Criteria Builder
A marking rubric for a task you set — criteria against bands, with descriptors that actually tell a 7 from a 5, the marks allotted to each criterion, and a note on what a top answer does.
This is a cloud tool. The task description and the options you set are sent to the model. No pupil work and no names go with it. Nothing is stored by us; the call is counted against your month’s allowance — 10 free, 300 on Pro. How the AI tools handle data.
Tips
- Paste the task exactly as the pupils saw it, including the mark count and what you told them would be assessed. A rubric written from the real question differentiates; a rubric written from "an essay about 1857" does not.
- Read down a single criterion, band by band, before you use it. That is where a weak rubric shows itself: if band 3 is band 4 with "some" in front of it, the descriptors do not differentiate and you should say so in the task box and run it again.
- The mark arithmetic and the missing-descriptor check at the bottom of the result are done here in your browser after the answer arrives, not claimed by the model. If the criteria do not add up to the total, you will be told in plain numbers.
- The borderline calls are the part to take to a department meeting. Agreeing three of those in advance does more for consistent marking than another page of descriptors.
- Name your framework only if you know what it says. Asked for "IB MYP criterion B" it will align to it as it understands it — check that against the real criterion before a moderated assessment, because a subject guide is not something it has in front of it.
Frequently asked questions
Can I use this for an official moderated assessment?
Not as it stands. For anything that counts towards a qualification the awarding body’s own criteria are the criteria, and no generated rubric can replace them. This is for your own assessments, your department’s common tasks, and for turning an official criterion into something pupils can read.
Why are criteria an object instead of a table?
Because a rubric read as a grid on paper is read as a list when you are marking: you take one criterion and work down its bands. The keyed shape puts each criterion’s four descriptors together where you can compare them, and the markdown download lays them out the same way.
Will pupils understand it?
The descriptors are written for a marker. Ask in the task box for a pupil-facing version — "also write each band in the second person for pupils" — and you get both, which is usually the version that improves the next piece of work.