Methods · Dialogue, argument & inquiry · Short guide
C25
Multi-critic panel
Receive simultaneous critique from several distinct critical personas and reconcile them: training judgment under conflicting advice, not obedience to the loudest voice.
When to use it
- You want conflicting advice on purpose, not one blended voice.
- An exam or review will hit your work from several angles.
- You tend to obey the loudest critique instead of reconciling them.
What to ask AI
Type something like: “Multi-critic panel on this work: [ ]. Critics stay in role and label every paragraph with their name.”
Choose distinct roles (methods stickler, clarity editor, devil's advocate…).
Require a reconciliation memo written by you.
How it helps
One critic produces one taste. A panel produces conflict.
Multi-critic panel trains judgment under conflicting advice: what do you keep, cut, or transform?
The reconciliation memo is the learning, not obedience to the loudest persona.
How to run it
- Paste a short piece of work.
- Name three distinct critic roles.
- Receive labelled critiques without merging them.
- Write a memo: what you accept, reject, and why.
- Revise once from the memo, not from raw panic.
Example exchange
Panel this abstract: [paste]. Roles: methods stickler, clarity editor, sceptical examiner. Label every comment.
Methods: sample size missing. Clarity: second sentence is a pile-up. Examiner: claim outruns evidence in line 3.
I’ll add n, split the sentence, and hedge line 3. I’m rejecting the clarity demand to cut the mechanism clause, it’s load-bearing.
Good, you reconciled the panel instead of obeying everyone.
Copyable prompt
Multi-critic panel on this work: [ ]. Critics (stay in role, label every
paragraph with your name): 1) [ROLE] 2) [ROLE] 3) [ROLE].
Do not merge advice. After critiques, wait for my reconciliation memo.
The Tell
If you can quote critics and cannot produce a memo of what you decided, you collected voices, not judgment.
If the AI merges the panel into one bland verdict, restart with labelled roles and a memo requirement.
Principle evidence
Strength of the underlying learning idea, not a claim about AI products.
The underlying learning idea is rated weak. Multi-persona critique panels are a design pattern for surfacing conflicting standards. Evidence is thinner than for simpler critique-and-revise loops; treat as practice in reconciling advice.
AI delivery evidence
Whether an AI tutor delivers this method well is a separate question.
AI personas can be stereotyped or contradictory without being useful. Your memo is the method. Delivery is speculative.