Artificial Intelligence

Short Briefing · Evidence current through 2026-09-22

How do I show AI the format I want?

Three lost-property notes arrive in different styles: a bottle by the desk, a scarf in a gallery, and a glove with no date recorded. Asking for a neat list leaves the format open to interpretation. Show one completed entry, then name the fields you want repeated. The example explains the pattern; explicit labels make it easier to check whether each new entry follows it.

For
general education, not made specifically for children
Use it to
Give a short example and explicit field labels.

Watch the briefing

Watch the full video on YouTube

Briefing

Three lost-property notes arrive in different styles: a bottle by the desk, a scarf in a gallery, and a glove with no date recorded. Asking for a neat list leaves the format open to interpretation. Show one completed entry, then name the fields you want repeated. The example explains the pattern; explicit labels make it easier to check whether each new entry follows it.

Our completed example is a blue umbrella found on October eighth in the lobby, stored in cabinet A. It has four fields: item, date found, location found, and storage. The labels separate where something was discovered from where staff put it afterward. That distinction matters when a visitor asks about an item. This is an invented record, with no personal information or real ownership claim.

Now give the three new source notes. A green bottle was found at the front desk on October ninth and placed in cabinet B. A red scarf was found in gallery two on October tenth and placed in cabinet A. One black glove was found in the courtyard and put in cabinet B, but its date was not recorded. No note names an owner. Those missing facts must not appear by invention.

Our request says: create three entries with exactly the same four labels as the example, in the same order. Copy only facts supported by each note. Use not recorded for a missing date, and add no owner or return status. The completed umbrella entry is a format example, not a fourth item to process. Separating the example from the new inputs helps us avoid counting or copying it as new work.

Our flawed illustrative output dates every entry October eighth, copying the example's value instead of its pattern. It also merges found location with storage. The corrected mock gives the bottle October ninth, front desk, cabinet B; the scarf October tenth, gallery two, cabinet A; and the glove a date of not recorded, courtyard, cabinet B. The format repeats. The evidence for each value stays with its own source note.

Review the structure and the facts independently. Do we have three entries, all four labels, and the requested order? Then check every value against the matching note. A perfectly shaped record can still contain the wrong date. A short example may also leave edge cases unclear, which is why we supplied an explicit missing-date rule. Clear prompting guides a draft; it does not certify the resulting records or establish ownership.

Give a short example and explicit field labels. Use realistic but shareable sample data, identify the new inputs separately, and state what to do when information is absent. Repeat the pattern, not the example's facts. Then inspect both format and content before using the entries. Next, we will handle a task where an unknown number prevents a responsible answer.

One insight you can use

Give a short example and explicit field labels.

What remains uncertain

The approved narration states the applicable limits; teaching examples are not measured outcomes or individualized recommendations.

Disclosures

  • AI-assisted production and synthetic narration. Original teaching examples and diagrams; linked third-party sources retain their respective rights.

Corrections

  • No corrections have been recorded.

Original sources and limits

See what supports the briefing

  1. https://developers.openai.com/api/docs/guides/prompt-engineeringdevelopers.openai.com · 2026-09-17

    First-party technical guidance supports clear boundaries, examples and relevant context. Fictional demonstrations and checklists are original editorial applications, not guaranteed outcomes.