Review edition · Complete text lesson. Platform screenshots are being prepared.
Suppose you could…
Prepare for difficult questions
Prepare grounded answers and explicitly identify questions the evidence cannot answer.
Prepare grounded answers and explicitly identify questions the evidence cannot answer.
Time is a practice estimate. Pause or return whenever you need.A useful little possibility.
Your goal is a usable work artifact, not a perfect-looking AI answer. Prepare grounded answers and explicitly identify questions the evidence cannot answer.
Let’s make a start.
Use a fresh conversation in an AI tool you are permitted to use. Copy the fictional sample below, then paste the starter prompt with it. No file upload, web search or paid integration is required.
1. Write the hardest question you expect.
2. Run the prompt.
3. Trace each factual answer to the sample.
4. Practice saying “we do not yet know” followed by an appropriate next check.
Your sample material
Fictional training material. No real customer information needed.
Proposal: a second workshop pilot costing €220. Evidence: 12 first-pilot attendees, 8 feedback responses, 6 positive on usefulness. No productivity measurement or comparison group. Venue access unconfirmed. A larger rollout is not approved. Presenter wants a decision on conditional pilot approval.Download sample .txt ↓
Generate six likely skeptical questions and concise answers using only the evidence. For each answer label supported, partially supported or unknown. Include questions about cost, benefit, representativeness, access, alternatives and the decision requested. Never invent facts to defend the proposal.
Keep personal and confidential information out of external AI tools. Start with the sample material.
Turn a first try into your thing.
Compare the result with the checklist before opening the worked example. Identify one specific weakness and ask for a targeted revision, referring to the original sample. If all checks pass, tighten the wording without changing the facts. Save the reviewed result in your own document; do not send or publish it as part of this exercise.
Pause for the “aha.”
AI can help rehearse challenge, but it should not become a machine for rationalizing your preferred conclusion. An unanswered question can be a useful result.
Useful beats convincing.
Check these against your actual output. Your answers stay on this device; this is self-review, not automated grading.
Finished your first try? Reveal a worked example ↓
Cost? Supported: proposed €220; no larger rollout cost is known. Benefit? Partially supported: 6 of 8 respondents found it useful, but productivity gains are unmeasured. Representative? Unknown: 8 responses from 12 attendees do not establish wider benefit. Accessible? Unknown: venue access needs confirmation. Alternatives? Unknown as evaluated options; deferring or changing venue can be proposed, not described as tested. What is being approved? Supported: a conditional second pilot, not full rollout. A strong answer makes the evidence boundary clear rather than sounding certain.
A new case. Your own approach.
Apply the review principles to these new facts, not the previous sample’s specific numbers or answers.
Prepare four skeptical questions about a proposed guide rewrite: five volunteers liked a draft; nobody tested task completion; cost estimate €80 is provisional. Answer without inventing a return on investment.
One possibility leads to another.
Repeat on a small, permitted task of your own. Record what you changed and which check caught a problem. Keep using fictional inputs if workplace material is not approved for your chosen tool.
Your reusable workflow card ↓
Task: Prepare for difficult questions Audience: [who will use this] Approved source material: [paste] Required output: Prepare grounded answers and explicitly identify questions the evidence cannot answer. Constraints and unknowns: [list] Review before use: - Six relevant questions are covered. - Every factual answer is supported by the sample. - Unknowns are admitted explicitly. - Possible alternatives are not called evaluated options. - The requested decision remains conditional pilot approval.Download workflow .txt ↓
You tried something. That counts.
Keep a note of what you made or what surprised you. It stays in this browser.