/ Home

Prompting

Note: Rules Matter!

🧪 Science of GenAI Prompting – Basics

🔹 What is Prompting in GenAI?

Prompting is the method of giving instructions to a Generative AI model (like ChatGPT) in the form of text, examples, or context so that it produces the desired output.
👉 In simple terms: Prompts are the questions or commands we give to GenAI.


🔹 Why Do We Need Prompting for GenAI?


🔹 Disadvantages if We Don’t Know Prompting Well


🔹 How Can We Learn Prompting?

  1. Practice – Try different prompt styles and compare results.
  2. Study Examples – Learn from proven prompt templates.
  3. Experimentation – Treat prompting like a scientific process: test → refine → repeat.
  4. Feedback – Share prompts with peers, improve collaboratively.

🔹 5-Step Framework of Prompt Engineering

  1. Task – Clearly define what you want the AI to do.
  2. Context – Provide background details to guide the response.
  3. References – Add examples, data, or style instructions for accuracy.
  4. Evaluate – Check the output for relevance, clarity, and correctness.
  5. Iterate – Refine the prompt step by step until you get the best result.

Nmeonics: Tiffen Coffee Ready, Eat Immediately


🔹 5-Step Framework of Prompt Engineering (Thalapathy Vijay Edition)

  1. Task – Define what you want.
    • Example: “Write a short movie review of Vijay’s Master in 5 lines.”
  2. Context – Provide background for better answers.
    • Example: “Assume the audience are college students who love mass scenes.”
  3. References – Add examples or style to guide output.
    • Example: “Here’s a style sample: ‘Ghilli is a thrilling sports-action film that mixes humor with action.’
      Now review Leo in the same style.”
  4. Evaluate – Check if the AI’s output matches what you need.
    • Example: Did it capture Vijay’s screen presence, dialogues, and fan moments?
  5. Iterate – Refine the prompt for improvements.
    • Example: If the review is too generic, update:
      “Focus on Vijay’s performance, action choreography, and music by Anirudh.”

🔹 Prompting Techniques

  1. Instructional Prompting – Direct and clear commands.
    • Example: “Summarize this article in 3 bullet points.”
  2. Role-based Prompting – Ask the AI to act in a role.
    • Example: “Act as a career coach and suggest IT jobs for freshers.”
  3. Chain-of-Thought Prompting – Ask AI to reason step by step.
    • Example: “Think step by step and explain how solar panels produce electricity.”
  4. Few-shot Prompting – Provide examples to guide output style.
    • Example: “Example: Apple: A fruit that is red and sweet. Mango: A fruit… Now describe Orange.”
  5. Prompt Chaining – Break a complex task into smaller prompts.
    • Example: First generate an outline → then expand each section → then refine tone.

Types of Prompting

🎬 Thalapathy Vijay – Prompting Examples

Here are real-time examples of the 4 main types of prompting using Thalapathy Vijay as the subject.


🔹 1. Instructional Prompt

👉 Direct, clear instruction.

Prompt:
List 3 of Thalapathy Vijay’s blockbuster movies in bullet points.

Expected Output:


🔹 2. Role-based Prompt

👉 Model acts with a persona.

Prompt:
You are a movie critic. Describe Thalapathy Vijay’s acting style in 3–4 sentences.

Expected Output:
Thalapathy Vijay is known for his charismatic screen presence and mass appeal.
His acting blends action, comedy, and strong emotional delivery.
He often portrays underdog heroes who rise against injustice, resonating with fans across generations.


🔹 3. Chain-of-Thought Prompt

👉 Step-by-step reasoning.

Prompt:
Think step by step. Why is Thalapathy Vijay called the ‘Ilayathalapathy’ and later ‘Thalapathy’? Explain clearly.

Expected Output:

  1. Vijay started his career as a child actor.
  2. His father, director S.A. Chandrasekhar, introduced him as a hero.
  3. Early fans called him ‘Ilayathalapathy’ (Young Commander).
  4. As his popularity grew, the ‘Ilaya’ (young) title was dropped.
  5. Today, he is simply known as ‘Thalapathy’ (Commander), symbolizing leadership.

🔹 4. Few-shot Prompt

👉 Provide examples, ask for similar output.

Prompt:
Examples:

Now describe Thalapathy Vijay in the same style.

Expected Output: