Prompt Engineering (Basics)

Prompt engineering is the practice of crafting inputs (called prompts) to get better and more accurate outputs from a generative AI model. Think of it as giving smart instructions to a very powerful assistant.

Even though the model is trained on vast data, how you ask still matters a lot.

🧠 What Is a Prompt?

A prompt is simply the input you give to a language model like ChatGPT or Claude.

  • It could be a question:

    "What are the benefits of meditation?"

  • A command:

    "Write a haiku about winter."

  • Or a system-style instruction:

    "You are a friendly customer support agent. Answer in short, clear sentences."

✨ Why Prompt Engineering Matters

Good prompt engineering helps you:

  • Get more accurate or creative answers

  • Avoid hallucinations or vague replies

  • Make the model behave in a specific tone or style

  • Guide the model to follow multi-step instructions

🔑 Basic Prompting Techniques

Technique
Example Prompt

Be Specific

"List 3 benefits of yoga for mental health."

Set the Role

"You are a career coach. Give resume advice."

Use Format Hints

"Respond in bullet points."

Ask Step-by-Step

"Explain how a car engine works, step by step."

Few-shot Prompting (show examples)

"Translate to French: Apple = Pomme, Car = Voiture, House = ?"

🧠 Prompt Example

Basic Prompt:

“Explain photosynthesis.” Output: “Photosynthesis is the process by which plants make food…”

Improved Prompt (engineered):

“You are a science teacher explaining photosynthesis to 10-year-olds. Use simple language and a fun example.” Better Output: “Photosynthesis is how plants make their own food using sunlight—kind of like how we cook using a stove!”

⚠️ Common Prompt Mistakes

  • Too vague: “Tell me something.” → Model guesses your intent

  • Too complex: “Summarize this, then write an essay, then list keywords.” → Confuses the model

  • Lacking format: “What are types of AI?” → One long paragraph instead of bullets

🧪 Practice Tip

Start with a basic prompt, then iterate:

  1. Add role

  2. Add style

  3. Ask for format

  4. Break into steps

Prompting is like programming — the clearer the instructions, the better the output.

PreviousKey Concepts: Tokens, Prompts, Context WindowNextChapter 2 - Models & Architectures

Last updated