Attachments
  • Mark as Completed
  • 17
  • More
Previous class
Section 4 : Iteration loops

Section 5: “Show, don’t tell” prompting

Purpose of This Section

This section explains why showing examples is often more effective than describing intentions when working with AI systems.

AI models are optimized to recognize patterns. Examples give the system a concrete target, reducing ambiguity and improving alignment.

The Core Idea

AI responds more reliably to demonstrations than to abstract descriptions.

When you only tell the AI what you want, it must interpret vague terms like “professional,” “clear,” or “better.” When you show the AI an example, you eliminate guesswork.

Examples make expectations visible.

Why Telling Often Fails

Abstract instructions rely on shared interpretation.

Words like “formal,” “friendly,” or “on-brand” mean different things to different people. When used alone, they force the AI to guess which interpretation you intend.

Guessing increases the risk of misalignment.

Why Showing Works

Examples provide concrete reference points.

They allow the AI to:

  • match tone and structure
  • replicate formatting patterns
  • align style and level of detail
  • avoid unwanted approaches

Showing replaces interpretation with pattern matching.

What Counts as Showing

Showing can include:

  • a paragraph you like
  • a sample email or document
  • a before-and-after comparison
  • an example of what to avoid

You do not need perfect examples. Even partial references improve results.

Show and Tell Together

Showing does not replace instructions.

The strongest prompts combine both:

  • instructions explain the task
  • examples demonstrate the outcome

Together, they create clarity without rigidity.

Common Failure Mode

A common mistake is over-explaining intent without providing reference material.

Users describe goals in detail but leave the AI to imagine what success looks like. This often leads to repeated revisions.

Showing shortens the loop.

The Conjugo Rule

If the output isn’t matching your expectations, show an example.

Concrete references outperform theoretical descriptions.

Section Takeaway

  • AI recognizes patterns better than intent
  • Examples reduce ambiguity
  • Showing accelerates alignment
  • Combining examples with instructions produces the best results

Showing is not extra work. It is a shortcut.

This concludes Section 5.