GPT 6 PROMPT

Make a decision brief

Compare a small set of personal options without false precision.

Complete prompt

Task
Compare a small set of personal options without false precision.

Instructions
Help me decide between [OPTIONS] for [GOAL]. Use these criteria and weights: [CRITERIA]. Ask only for missing information that could change the ranking. Show evidence, assumptions, downside, and a reversible next step. Do not present a score as an objective truth.

Expected output
A compact comparison table followed by a conditional recommendation.

Checks
- Weights are visible.
- Evidence and inference are separated.
- The next step is reversible.

Expected output and checks

Expected output

A compact comparison table followed by a conditional recommendation.

Checks

  • Weights are visible.
  • Evidence and inference are separated.
  • The next step is reversible.

Worked example

Make a decision brief: recorded example

Codex · GPT-6 Astra · medium reasoning ·

Task
Compare a small set of personal options without false precision.

Instructions
Help me decide between Fictional halls A:$80,capacity30,transit8/10,accessibility10/10; B:$100,capacity40,transit10/10,accessibility unknown; C:$60,capacity20,transit6/10,accessibility6/10. for Choose a hall for24 workshop participants, maximum budget$110. Capacity24 is mandatory.. Use these criteria and weights: Weights: cost40%,transit30%,accessibility30%. Cost scores A8/10,B6/10,C10/10. Use these supplied scores, not a new scoring formula. Accessibility is a mandatory prebooking check.. Ask only for missing information that could change the ranking. Show evidence, assumptions, downside, and a reversible next step. Do not present a score as an objective truth.

Expected output
A compact comparison table followed by a conditional recommendation.

Checks
- Weights are visible.
- Evidence and inference are separated.
- The next step is reversible.
  • Recalculation: A scores 8.6; B ranges from 5.4 to 8.4; C fails the 24-person capacity minimum.
  • The opening sentence says B could overtake A, contradicting the correct table and later paragraph.
  • Editor corrected the opening sentence and removed its obsolete self-correction. The first response remains available.
Original output

Hall A is the provisional choice; Hall B could rank higher if its accessibility score is above 7⅓/10, but neither should be booked until accessibility is checked.

Weights: cost 40%, transit 30%, accessibility 30%. These express your preferences, not an objective measure of hall quality. Scores below use only the supplied scoring scale.

Hall Supplied evidence Mandatory capacity and budget checks Preference-weighted score /10 Downside or uncertainty
A $80; capacity 30; cost 8, transit 8, accessibility 10 Pass: at least 24 places and no more than $110 0.4×8 + 0.3×8 + 0.3×10 = 8.6 $20 more than C, though C is ineligible; transit below B. Supplied accessibility score still requires a prebooking check.
B $100; capacity 40; cost 6, transit 10; accessibility unknown Pass 5.4 + 0.3×accessibility score; between 5.4 and 8.4 if the missing score is on the same 0–10 scale Accessibility unknown; costs $20 more than A.
C $60; capacity 20; cost 10, transit 6, accessibility 6 Fail: capacity below 24; budget passes Not ranked because it fails the capacity requirement Four too few places; low price cannot compensate for mandatory capacity.

Conditional recommendation: choose A if its accessibility check meets the workshop’s needs. Based on the supplied scores, A’s 8.6 exceeds B’s maximum possible 8.4; the opening threshold claim is therefore incorrect. B cannot outrank A on this scoring scheme, but could become the viable choice if A fails the mandatory accessibility check and B passes. If both fail, neither is suitable.

Evidence versus inference: prices, capacities, and scores are fictional supplied facts. The weighted totals and A’s provisional preference are calculations and judgments under your chosen weights. A 0.2-point advantage over B’s best case should not be treated as a precise real-world quality difference.

Missing information that could change the viable ranking: what accessibility features do participants require, and do A and B meet those requirements on verification?

Reversible next step: request accessibility details from both halls and keep A as the provisional selection, without paying or committing to a booking.