Reduce the Precision

Release mode Beta

Level
Zero / Beginner
Format
individual, pair, or small group

Practice the principle of using the least precise information that still supports a task. Precision is not always wrong; it should earn its place. Learners change identifying details into useful categories, approximations, or fictional equivalents while preserving the goal of the request.

Keep a request useful without making a person or organization easier to identify

What you will make

Practice the principle of using the least precise information that still supports a task. Precision is not always wrong; it should earn its place. Learners change identifying details into useful categories, approximations, or fictional equivalents while preserving the goal of the request.

By the end, you can recognize when a detail is more exact than a task needs, choose a safer replacement, explain what the rewrite preserves and loses, and flag a situation that should be handled outside a general chat request.

Lab materials

Use during the lab

Go deeper

Before you start

Level and format: Zero / Beginner; individual, pair, or small group

You need: rewrite cards, paper or a text editor, and a chat tool only if a facilitator chooses an optional comparison. Use fictional cards only; do not bring real names, addresses, accounts, medical records, employee details, school records, client material, or private messages into the activity.

The core idea

Useful context is not the same as maximum detail. This lab reduces the precision of what a tool receives, not the precision of the finished artifact. A final reminder may need an exact date, time, or address even when the drafting request uses placeholders that a person fills in afterward.

A request may need an audience, a broad situation, a constraint, or a verified fact. It rarely needs every exact identifier that happens to be available. A private detail may legitimately shape how you ask—for example, making an offer easy to accept without explanation—without becoming what you send. Reducing precision does not create permission or authority, and it does not make a sensitive task automatically suitable for a chatbot or account; sometimes the right choice is to stop, remove the task, or use an approved human process.

Keep enough detail without oversharing

Complete the activity here. Open the optional Lab Notebook beside this page when you want a place to record your work.

  1. Choose a Card

    Choose one fictional rewrite card. Underline the actual task. Circle the details that help the task and mark details that could identify a person, household, organization, account, or private situation.

  2. Ask What the Task Needs

    For each marked detail, ask:

    • Would the request still work without this exact detail?
    • Could a range, role, category, region, or fictional example do the same job?
    • Whose information is this, and am I authorized to use it for this purpose?
    • Does the task need a human or approved system rather than a general chat request?
    • What would become inaccurate or less useful if the detail changed?

  3. Make a Safer Rewrite

    Rewrite the card’s request so it keeps the goal and useful constraints while removing unnecessary precision. You may:

    • replace an exact value with a useful range or category;
    • replace a name or organization with a role or fictional equivalent;
    • summarize a private situation without reproducing it;
    • use a placeholder, then have an authorized person fill in the exact value afterward;
    • move a record-specific action outside the request; or
    • state an uncertainty instead of filling it in.

    Do not merely remove every fact. A good rewrite still gives a later reader enough context to understand the requested task.

  4. Compare

    Exchange rewrites with a partner or compare them against the original. Identify:

    • one detail that remained because it was necessary;
    • one detail that became less precise;
    • one detail that stayed out entirely; and
    • one tradeoff in usefulness, accuracy, or safety.

    Before moving to the optional tool check, record one boundary that a safer rewrite cannot solve: who must decide, verify, supply an exact value, or use an approved process.

  5. Optional Tool Check

    If a facilitator permits an approved tool, use only the fictional rewrite. Ask for the bounded task, then check whether the response invents names, facts, policy, or a level of certainty you did not provide. Record the optional result separately from the human or approved-process boundary you already named. A tool response is optional; the learning happens in the rewrite.

Check your result

  • Compare your record with the source and the stated task. Name what the result supports, what remains uncertain, and what must be checked by a person or authoritative source before anyone acts.

  • A polished result is not proof that it is accurate, complete, safe, or appropriate. Keep unresolved questions visible and verify consequential claims outside the tool.

You are done when

What did the task actually need? What did the original detail add besides identifiability? When would reducing precision be insufficient because the task itself needs a different process?

Done means you can show a before-and-after request, name one useful detail you retained, name one precision change you made, and explain one remaining human or policy boundary.

Try another version

These are optional extensions, not additional completion requirements.

Learner Map

Tracks

Modules

Help us improve this Activity

Tell us what helped, what slowed you down, or what you needed but could not find.

🎓 Give learner feedback