Chapter 11: Prompt Engineering Basics

“Talk to models like you mean it.”

This chapter is one of the most human in this journey — it’s not about code, or containers, or GPUs. It’s about language. And how we, as builders, can shape AI behavior simply by choosing the right words.

Welcome to the artful world of Prompt Engineering.

---

This Chapter Covers

• What prompt engineering is (and why it matters)
• The structure of a good prompt
• Task types: classify, generate, reason, summarize
• Techniques: instruction, examples, tone, constraints
• Builder’s lens: shaping intelligence through intention

---

Opening Reflection: Words as Levers

“Give me the right words, and I will move the model.”

You don’t need to change the architecture. You don’t need to retrain the weights. You don’t even need to touch the API.

You just need to say the right thing — in the right way — and watch the model become:

• A poet
• A helper
• A data analyst
• A sarcastic comedian
• A highly specific meme captionist

This is prompt engineering: The art of talking to machines… and getting exactly what you meant.

---

11.1 What Is Prompt Engineering?

Prompt engineering is the craft of designing inputs to language models (e.g. GPT, Claude, Mistral) that guide the model to produce useful, accurate, or creative output. It’s part programming, part psychology, part UX.

---

11.2 Prompt Structure: The Core Recipe

Instruction  
(Optional) Examples  
Constraints / Formatting  
User Input

Example (Sentiment Analysis Prompt):

Classify the sentiment of the following sentence as Positive, Negative, or Neutral: Input: "I don’t love this product, but it works." Sentiment:

The model fills in the blank: "Neutral"

---

11.3 Common Prompting Tasks

Task Type	Goal	Example Prompt
Classification	Label text (e.g., intent)	“Label this text as Happy, Sad, or Angry.”
Summarization	Compress info	“Summarize this article in 3 bullets.”
Generation	Produce text	“Write a tweet about AI in 10 words.”
Reasoning	Chain of logic	“Explain why gravity decreases with distance.”
Extraction	Pull structure from chaos	“Extract dates and names from this text.”

---

11.4 Techniques That Boost Prompt Quality

Technique	Example / Effect
Explicit roles	“You are a meme caption generator…”
Few-shot learning	Show 1–3 examples before user input
Chain of Thought	“Let’s think step by step…”
Output formatting	“Respond in JSON: {‘label’: … }”
Style injection	“Respond as if you're Shakespeare.”

---

11.5 Prompt Engineering in Code (OpenAI)

import openai

response = openai.ChatCompletion.create(
  model="gpt-3.5-turbo",
  messages=[
    {"role": "system", "content": "You are a product description writer."},
    {"role": "user", "content": "Describe a smart water bottle for athletes."}
  ]
)

print(response["choices"][0]["message"]["content"])

This is where the magic happens.

---

11.6 When Prompting Fails: Debug Like a Builder

Symptom	Try This Fix
Too generic / vague	Add examples or clarify instruction
Output too long/short	Add constraints: “<30 words” or “3 lines”
Repeats itself	Add: “Do not repeat yourself.”
Hallucinates info	Add: “Only use the info provided.”

Prompting is iterative. You’ll get better through play.

---

11.7 Builder’s Perspective: Your Prompt Is a Prototype

“In a world where models know everything, what matters is how you ask the question.”

Your prompt:

• Is your interface
• Is your architecture
• Is your business logic
• Is your UX

It's the single string of text that determines whether your app feels:

• Confident
• Helpful
• Funny
• Human

You don’t need to know more. You need to ask better.

---

Summary Takeaways

Concept	Why It Matters
Prompting = shaping behavior	No code changes needed
Clear, specific input	More reliable, useful outputs
Few-shot + structure help	Reduces hallucination, increases control
Prompt = soft interface	Easy to change, test, and improve

---

🌟 Closing Reflection

“The model has intelligence. You have intention. Prompt engineering is the conversation between the two.”

---