Discussion Prompt Redesign Checklist
Test how AI-vulnerable a discussion prompt is, then redesign it with four levers while keeping its learning goal.
By Linda Kowalcky, CSSH Director of AI Curricular Transformation. See it applied in three weeks of redesigned public policy prompts.
Step 1 · Diagnose: is this prompt AI-vulnerable?
- Could a generic AI tool answer it well without having done the reading?
- Could a student answer from general or news knowledge instead of the specific text?
- Does the prompt only ask “what do you think?” or “what resonates with you?” about a general concept?
If yes to any of these, use one or more of the levers below.
Step 2 · Apply the Four Levers
Generic: “Which principle resonates with you?”Anchored: “Name a specific policy you interact with (at work, school, home). Which principle explains why it succeeds or fails?”
Generic: “How did the policy change over time?”Traced: “Pick one specific design element that changed. Using [framework], explain why that element was the one that had to change.”
Addition: “Compare your examples with your group. Pick the two that best illustrate the same principle in different domains, and report those out.”
Addition: “Include one direct reference to [author/concept], page number or term, in your initial post before your group discusses.”
If an anchor asks about students' own lives, offer a non-personal option too.
Step 3 · Redesign worksheet
Fill this in for each prompt you revise.
| Question | Your answer |
|---|---|
| What is the core learning goal of this prompt? | |
| Is it currently vulnerable? (yes/no, and why) | |
| What specific or local referent can I anchor it to? | |
| What “why” or mechanism question can I add? | |
| What group-dependency step can I build in? | |
| What textual reference should I require? |
Step 4 · Sanity check
- Does the redesigned prompt still target the same learning objective as the original?
- Is it still answerable in the time students have?
- Have I added specificity, not just length or complexity?
Use the checklist each time you revise a prompt. Over time the four levers become second nature, and you'll start writing AI-resilient prompts from scratch.