Methods · Assessment & diagnosis · Short guide

H5

Rubric generation

Build the marking criteria before you attempt the task, so you know what “good” means while there is still time to aim at it.

When to use this
Before you start studying
Learning impact
Clear help, modest gain
Learning evidence
Credible with caveats
Where you are with the topic
Checking readiness
How often to use it
Useful once for a topic
What AI is doingCreate practice, examples, or schedules

When to use it

  • You have the assignment brief and no clear sense of what will actually be marked.
  • You usually reverse-engineer the rubric from the grade after submitting.
  • An official rubric exists, but the language is abstract and you need it in your own words.

What to ask AI

Paste the assignment and ask for a draft rubric: four to six dimensions with brief level descriptors.

Ask it to make you prioritise weights, do not let every dimension look equally important.

If you have an official rubric, paste it and force a diff: warn where you are ignoring official dimensions.

Forbid it from writing the assignment for you.

How it helps

Students often reverse-engineer rubrics from grades. Criteria first turns assessment into a target.

Even when an official rubric exists, rewriting it in your words and weighting what matters is the method.

Keep the rubric visible while you draft. Then self-mark before you submit.

How to run it

  1. Get the assignment prompt in front of you.
  2. Draft criteria (dimensions and levels) before writing.
  3. Compare to the official rubric if any; reconcile.
  4. Weight the top three dimensions.
  5. Keep the rubric visible while drafting, then self-mark before submit.

Example exchange

You

Assignment is a 1500-word policy brief. Help me draft a five-dimension rubric. I'll set the weights.

AI

Draft dimensions: claim clarity, evidence quality, counter-argument, structure, audience fit. Rank weights from one to five before I refine the descriptors.

You

Evidence 5, counter-argument 4, claim 3, structure 2, audience 1. Here's the official rubric, where do I diverge?

AI

You under-weighted policy feasibility, which is an official dimension. Add it, or accept you're aiming at a different target.

Copyable prompt

Assignment: [ ].
Help me draft a rubric: 4-6 dimensions, brief level descriptors. Ask me to
prioritise weights. If I paste the official rubric, force a diff and warn where
I'm ignoring official dims. Do not write the assignment.

The Tell

Feedback surprises you on dimensions you never considered.

That usually means the rubric arrived after the draft, or never at all.

Vibes criteria (“sounds smart”) also fail the method: force observable descriptors.

Principle evidence

Strength of the underlying learning idea, not a claim about AI products.

The underlying learning idea is rated moderate. Building rubrics before attempting clarifies goals (criterion-referenced assessment and assessment literacy). It is a clarity and self-regulation aid more than a magic grade booster.

AI delivery evidence

Whether an AI tutor delivers this method well is a separate question.

AI rubric quality and alignment to real mark schemes are speculative. Prefer diffing against your institution’s official criteria whenever they exist.

Related methods