Skip to lesson content
Free course

AI Foundations / Module 4 / M04-U09 · 7-9 minutes, including the knowledge check

M04-U09 · 7-9 minutes, including the knowledge check

Module 4 Summary: A Prompt Is Not a Magic Spell

Consolidate the module decisions and keep the reusable resource available for later work.

What you can do now

You can turn a vague request into a task specification that is easier to produce, review, and correct.

The specification fields

  1. Outcome: What deliverable and change are required?
  2. Audience: Who will use it, and what do they know?
  3. Context: What background changes how the task should be done?
  4. Sources: Which authorised material may support the result?
  5. Constraints and boundaries: What must, must not, and cannot happen?
  6. Format: What structure makes the result useful and inspectable?
  7. Review criteria: How will a person judge the result?
  8. Workflow and authority: Which stages, checks, owners, and approvals are required?
  9. Stop conditions: When should the work pause or change method?

Main lessons

  • A topic is not an outcome.
  • More context is useful only when it is relevant and authorised.
  • Examples show patterns but are not automatically factual sources.
  • Constraints should prevent real failures, not decorate the prompt.
  • Observable criteria are stronger than words such as excellent or perfect.
  • Complex work needs stages that preserve human decisions.
  • Diagnose the defect before revising the specification.
  • Clear instructions improve work but do not guarantee correct output.

Before the knowledge check

The check has ten questions. You need 8 out of 10. Read the feedback before trying again.

Use the reusable Task Specification Template for future course activities and suitable real work.

What comes next

A good specification makes an output easier to inspect. It does not prove the output is accurate or complete.

Module 5 teaches a systematic review: what to check, how to verify material claims, and how to record an accept, revise, verify, or reject decision.

Keep this rule:

Improve the description of the work, then check the result against evidence.