Skip to content

Artificial Intelligence · Evidence and source dates listed below

How do I get useful alternatives instead of near copies?

Define the axis on which options should differ.

Watch the briefing

Watch the full video on YouTube

Full briefing transcript

You ask for three library display ideas and receive three versions of books on a table. The names differ, but the decision does not. To get useful alternatives, name the dimensions that should change. For this fictional display, we will compare cost, space, and setup time. Then require identical fields for every option, so the differences are visible rather than hidden behind enthusiastic descriptions.

Our library wants an adult display promoting local walking books. We have three invented proposals: table cards, a freestanding board, and a two-table trail. The figures are teaching estimates, not supplier quotes or measured library results. No materials are purchased and nothing is installed. Each plan must show materials cost, footprint, setup time, and a trade-off. We have not decided which dimension matters most yet.

The prompt asks for alternatives that differ on those dimensions while using the supplied estimates. Table cards cost eight currency units, occupy one point two by zero point six metres, and take twenty minutes to set up. The board costs twenty four, occupies one by zero point five metres, and takes forty five minutes. The two-table trail costs twelve, occupies two point four by zero point six metres, and takes thirty minutes.

These are illustrative options we wrote for the lesson, not recorded AI results. Table cards are compact and quick but show fewer covers. The board is vertical and needs an approved board already available. The trail offers more browsing space but uses more floor area. Reveal one row at a time, then highlight a single comparison column. The visual should help a viewer compare, not ask them to read three paragraphs at once.

Check the numbers before choosing. The display footprints are zero point seven two, zero point five, and one point four four square metres. Those figures exclude circulation and accessibility clearance; they are not an approved room layout. Table cards are cheapest and quickest in our estimates. The board has the smallest display footprint but costs more and takes longer. There is no universal best choice without a stated priority and real-world constraints.

If the library has little setup time, table cards may be worth examining first. If floor space is the main issue, investigate the board and its practical requirements. Ask whether equipment is available, estimates are realistic, and the layout meets local accessibility rules. Official guidance supports explicit requirements and criteria; our comparison is an editorial planning method. Different-looking proposals do not establish feasibility, and an assistant should not silently invent a ranking rule.

Define the axis on which options should differ. Request common fields, then compare the dimensions that actually influence your decision. Our three library ideas differ in inspectable ways, but the librarian still verifies costs, space, equipment, and suitability before acting. A useful choice is more than a clever title. Next, we ask an assistant to find what an apparently complete plan forgot.

Practical takeaway

Define the axis on which options should differ.

Disclosures

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

Original sources and limits

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

    Explicit instructions and context; outputs can vary. Developer guidance adapted to ordinary text workflows, not evidence these examples work reliably.

  2. https://developers.openai.com/api/docs/guides/evaluation-best-practices 2026-09-17

    Defined objectives and checks; technical guidance adapted to human review, not certification of the original worksheet.