Purpose
Use this lesson to compare options against weighted criteria without hiding judgment. The aim is to produce work that a colleague can review and act on, not merely text that sounds convincing.
Lesson Explanation
Scenario: Choose between three help-desk platforms.
Better instruction: Build a decision matrix using cost, integration, security, ease of use, and support. Ask me for weights before scoring; show the score calculation and any assumptions.
A decision matrix makes trade-offs visible, but it does not create objective truth. Define criteria, weights, scoring scale, evidence source, and sensitivity checks. The model should not invent weights or vendor facts. Weights are requested, scoring is visible, and assumptions are stated.
Work in a short cycle: provide the relevant input, state the result and limits, inspect the draft against the acceptance criteria, then revise the instruction when a requirement is missing. Keep the final human decision with the person responsible for the work.
Practice Questions
-
A prompt asks an AI to “score these three options and tell us which is best” without asking the AI to request weighting first. What does this lesson recommend instead?
View Answer
Asking for weights before scoring, since a decision matrix requires defined criteria and weights, not an AI-invented ranking.
-
What does this lesson say a decision matrix does NOT create, despite making trade-offs visible?
View Answer
Objective truth – a decision matrix makes trade-offs visible, but the underlying weighting and scoring are still choices, not objective facts.
-
Write a prompt for a decision matrix comparing three cloud storage providers, following this lesson’s guidance.
View Answer
Something like: “Build a decision matrix using cost, security, ease of migration, and customer support. Ask me for weights before scoring; show the score calculation and any assumptions.”
-
A decision matrix is generated with weights the AI invented on its own, without asking the requester first. What does this lesson say about this practice?
View Answer
The model should not invent weights or vendor facts – weights should be explicitly requested from the person making the decision.
-
What five elements does this lesson say a decision matrix should define?
View Answer
Criteria, weights, scoring scale, evidence source, and sensitivity checks.
-
A completed decision matrix shows the requested weights, a visible scoring calculation for each option, and explicitly stated assumptions. What does this reflect?
View Answer
A properly built decision matrix following this lesson’s full guidance.
-
A decision matrix produces a final ranking with no visible scoring math behind it. What does this lesson say is missing?
View Answer
Visible scoring – the score calculation should be shown, not just a final ranking presented without its underlying math.
-
Why does this lesson caution that “a fluent claim is not automatically a lawful or true claim” in the context of a decision matrix?
View Answer
A confidently stated fact used to score an option (like a vendor claim) might not actually be verified, so evidence sources for such facts need to be checked, not simply trusted.
-
A decision matrix compares software vendors using invented cost figures the AI wasn’t given by the user. What element of this lesson’s guidance does this violate?
View Answer
The AI should not invent vendor facts – actual, verified data should be provided rather than assumed.
-
What does this lesson recommend as a way to test how sensitive the final decision is to changes in weighting?
View Answer
A sensitivity check.
-
A decision matrix is built with pre-set weights the requester never approved, and it produces a recommendation the requester disagrees with. What does this lesson suggest could have avoided this mismatch?
View Answer
Asking for weights before scoring, so the criteria importance reflects the actual decision-maker’s priorities rather than an assumed default.
-
A prompt requests a decision matrix and explicitly asks the AI to state where the evidence for each score came from. What element of this lesson’s five components does this reflect?
View Answer
Evidence source.