Prompt Engineering9 min read

What is Prompting?

Talk to AI like a pro — the art of asking the right question
scope:Foundationaldifficulty:Beginner

Ordering Food in a Foreign Country

Imagine you're in a restaurant in a country where you barely speak the language. You're starving. You have two options:

Option A: You point vaguely at the menu and say "Food, please."

Option B: You carefully say "I'd like a medium-rare ribeye steak with mashed potatoes and steamed broccoli, no butter on the vegetables."

With Option A, you might get anything — sushi, soup, a salad, or something you're allergic to. With Option B, you get exactly what you want.

Talking to AI works the exact same way. The instructions you give to an AI are called a prompt, and the quality of what you get back depends almost entirely on how well you write that prompt.

What Exactly is a Prompt?

A prompt is simply the text you type into an AI system. That's it. Every message you send to ChatGPT, Claude, Gemini, or any other AI — that's a prompt.

But here's what most people don't realize: the AI doesn't "understand" your intent the way a human would. It doesn't read between the lines. It doesn't know what you meant to ask. It only has the exact words you typed.

This means:

  • A vague prompt gets a vague answer
  • An ambiguous prompt gets a random interpretation
  • A specific, clear prompt gets a specific, useful answer

Good Prompts vs Bad Prompts

Let's see the difference with real examples:

  • Bad: Tell me about Python — Are you asking about the snake or the programming language? A 5-minute overview or a PhD-level deep dive?
  • Good: Explain Python list comprehensions with 3 simple examples. Assume I know basic Python but am new to comprehensions.
  • Bad: Write an email — To whom? About what? What tone? How long?
  • Good: Write a polite 3-sentence email to my professor requesting a one-week extension on the homework due Friday. Mention that I've been sick.
  • Bad: Help with my code — What code? What language? What's the problem?
  • Good: My Python function is supposed to return the largest number in a list but returns None for empty lists. Here's the code: [paste code]. How do I fix it?
Note: The Golden Rule of Prompting: If a smart human would need to ask you a follow-up question to understand your request, your prompt isn't specific enough. Give the AI all the context it needs in the first message.

The Five Elements of a Great Prompt

Great prompts usually include some combination of these elements:

  • Task — What do you want the AI to do? ("Explain", "Write", "Compare", "Debug", "Translate")
  • Context — What background information does the AI need? ("I'm a beginner", "This is for a business presentation", "The audience is 10-year-olds")
  • Format — How should the output look? ("As a bullet list", "In a table", "In 3 paragraphs", "As Python code")
  • Tone — What style should the response have? ("Professional", "Casual", "Funny", "Academic")
  • Constraints — What limits or rules apply? ("Under 100 words", "Don't use technical jargon", "Include exactly 5 examples")

You don't need all five every time. But the more you include, the better your results.

The Magic of Being Specific

Watch how adding specificity transforms results:

  • Level 1: Write a story → generic, bland, unpredictable
  • Level 2: Write a short story about a dog → better, but still vague
  • Level 3: Write a 200-word children's story about a golden retriever named Biscuit who is afraid of thunder but discovers that thunder is just clouds clapping → specific, creative, delightful

Each level of specificity gives the AI more to work with. It's not about writing long prompts — it's about writing clear prompts with the right details.

Prompt Engineering in Practice

# Demonstrating how prompt quality affects AI output
# (Using a hypothetical AI function for illustration)


def simulate_ai_response(prompt):
    """Simulate how AI quality scales with prompt quality."""
    # Count specificity signals in the prompt
    specificity = 0
    signals = [
        'example', 'format', 'tone', 'audience', 'length',
        'step', 'list', 'table', 'beginner', 'expert',
        'don\'t', 'include', 'avoid', 'exactly', 'assume'
    ]
    for signal in signals:
        if signal.lower() in prompt.lower():
            specificity += 1
    
    # Rate the prompt
    if specificity == 0:
        quality = "Vague — AI will guess what you want"
    elif specificity <= 2:
        quality = "Decent — AI has some direction"
    elif specificity <= 4:
        quality = "Good — AI knows what you need"
    else:
        quality = "Excellent — AI can nail this"
    
    return quality, specificity


# Compare prompts
prompts = [
    "Tell me about sorting",
    "Explain bubble sort with an example",
    "Explain bubble sort step by step for a beginner. "
    "Include a Python example with a list of 5 numbers. "
    "Format the explanation as a numbered list.",
]


print("=== Prompt Quality Analyzer ===")
for i, prompt in enumerate(prompts, 1):
    quality, score = simulate_ai_response(prompt)
    print(f"\nPrompt {i}: \"{prompt[:60]}{'...' if len(prompt) > 60 else ''}\"")
    print(f"  Specificity signals: {score}")
    print(f"  Rating: {quality}")


# The five elements of a great prompt
print("\n=== Building a Great Prompt ===")
elements = {
    "Task":        "Explain bubble sort",
    "Context":     "for a beginner programmer",
    "Format":      "as a numbered step-by-step list",
    "Tone":        "friendly and encouraging",
    "Constraints":  "under 200 words, include one code example",
}


full_prompt = ". ".join(f"{v}" for v in elements.values())
print("\nElement breakdown:")
for element, text in elements.items():
    print(f"  {element:>12}: {text}")
print(f"\nCombined prompt: \"{full_prompt}.\"")
Output
=== Prompt Quality Analyzer ===

Prompt 1: "Tell me about sorting"
  Specificity signals: 0
  Rating: Vague — AI will guess what you want

Prompt 2: "Explain bubble sort with an example"
  Specificity signals: 1
  Rating: Decent — AI has some direction

Prompt 3: "Explain bubble sort step by step for a beginner. Include ..."
  Specificity signals: 5
  Rating: Excellent — AI can nail this

=== Building a Great Prompt ===

Element breakdown:
          Task: Explain bubble sort
       Context: for a beginner programmer
        Format: as a numbered step-by-step list
          Tone: friendly and encouraging
   Constraints: under 200 words, include one code example

Combined prompt: "Explain bubble sort. for a beginner programmer. as a numbered step-by-step list. friendly and encouraging. under 200 words, include one code example."

Iteration: Your Secret Weapon

Here's something that surprises people: great prompts rarely come out perfect the first time. The best AI users don't write one prompt and accept whatever comes back. They iterate.

The process looks like this:

  • Step 1: Write your best prompt and send it
  • Step 2: Look at what the AI gave you. What's good? What's wrong?
  • Step 3: Refine your prompt based on what was missing. "That's good, but make it shorter" or "Use simpler words" or "Add more examples"
  • Step 4: Repeat until you get what you need

This is how professionals use AI. They don't expect perfection on the first try. They have a conversation, guiding the AI closer and closer to what they want.

Think of it like giving directions. If someone ends up on the wrong street, you don't start over — you say "Turn around, go two blocks east." Same with AI. Build on what it gave you.

Challenge

Quick check

What is a 'prompt' in the context of AI?

Continue reading