Learn Better With AI

Methods · Assessment & diagnosis · Short guide

H10

Blind self-marking

Mark your own work against the rubric before seeing any feedback, catching gaps while you are still the first examiner.

When to use this
When checking yourself
Learning impact
Strong change
Learning evidence
Credible with caveats
Where you are with the topic
Checking readiness
How often to use it
Worth repeating often
What AI is doingEvaluate your work

When to use it

  • You finished a draft and usually jump straight to asking “what grade is this?”
  • External feedback always arrives after you have left the problem.
  • You want calibration: your scores versus an independent read.

What to ask AI

Paste the rubric and say you will self-mark first. Forbid comments until you paste scores and evidence quotes.

Then ask it to compare: where you were harsh or lenient versus its independent read.

Tell it not to overwrite your judgment, calibrate it.

How it helps

External feedback arrives after you have left the problem. Blind self-marking forces criteria application while revision is still possible.

Scoring each dimension with evidence quotes from your text is the work. Scores without quotes are vibes.

Later, comparing your blind scores to real or simulated marks builds calibration.

How to run it

  1. Finish a draft.
  2. Put away tips and prior AI chat.
  3. Score each rubric dimension with evidence quotes from your text.
  4. List fixes before any external comments.
  5. Then get feedback and compare, revise from the gaps.

Example exchange

You

Here's the rubric. I'm going to mark my own work against it first, please wait until I paste scores.

AI

Got it. When you're ready, send each dimension score with one quote that justifies it.

You

Evidence 3 out of 5. I wrote “Smith (2019) shows…” Counter-argument 2 out of 5. I never named a limit.

AI

I'd put evidence closer to 2: that quote cites, it doesn't weigh. You were lenient there and a bit harsh on structure. Calibrate those two before you revise.

Copyable prompt

Rubric: [ ].
I will self-mark my work blindly first. Do NOT comment until I paste my scores
and evidence. Then compare: where I was harsh/lenient vs your independent read.
Don't overwrite my judgment: calibrate it.
MY WORK: [ ]
MY SELF-SCORES: [after I write them]

The Tell

Your self-score is always “distinction” until the real mark lands.

Blind marking without evidence quotes is just optimism with a spreadsheet.

Self-marking in a chat already full of hints also fails, clear the hints first.

Principle evidence

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

The underlying learning idea is rated moderate. Marking one’s own work against a rubric before seeing feedback trains self-evaluation and reduces anchoring on an external mark.

AI delivery evidence

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

AI rubric application afterward is speculative. Use it to compare harsh/lenient patterns, not as the true grade.

Related methods