University

Hands-On Exercises in Prompt Engineering

Written by Aimee Bottington | Sep 23, 2024 2:23:17 AM

This lesson takes a practical approach to learning prompt engineering by offering hands-on exercises. You will work through guided scenarios where you craft, test, and iterate on prompts, with the goal of improving the quality of AI outputs. These exercises are designed to help you apply the principles discussed in previous lessons and refine your skills. By analyzing results and making adjustments, you’ll see firsthand how small changes in prompts can significantly affect AI behavior. Additionally, opportunities for peer feedback will allow you to compare strategies and improve your prompt engineering techniques.

Exercise 1: Refining Prompts for Creative Writing

In creative applications like writing stories, articles, or marketing content, crafting prompts that guide the AI while allowing creative freedom is essential.

  1. Task:
    Start with a broad prompt and refine it to create a well-structured, detailed output.

    Initial Prompt:

    • “Write a short story about a robot.”

    Steps:

    • Step 1: Analyze the initial output. Was the story engaging? Did it have a clear structure?
    • Step 2: Add details to guide the AI’s narrative.
      Refined Prompt:
      • “Write a 500-word science fiction story about a robot discovering human emotions on a distant planet. Include dialogue and vivid descriptions of the environment.”

    Outcome:
    By introducing constraints (word count, inclusion of dialogue) and specifying the story’s theme (robot discovering emotions), you guide the AI toward a more creative and structured result.

  2. Reflection:

    • Compare the outputs from both prompts.
    • How did adding context and constraints change the result? Did it improve the quality and focus of the story?

Exercise 2: Improving Customer Service Responses

In customer service, prompt engineering is crucial for ensuring that AI systems like chatbots deliver accurate and empathetic responses.

  1. Task:
    Develop a prompt for handling customer complaints and refine it to improve empathy and clarity.

    Initial Prompt:

    • “Help a customer resolve a billing issue.”

    Steps:

    • Step 1: Test the initial output to see how the AI handles the issue.
    • Step 2: Add specific instructions for tone and response structure.
      Refined Prompt:
      • “Use a polite and empathetic tone to help a customer resolve a billing issue. First, apologize for the inconvenience, then verify the billing details, and offer a solution. Ensure the response remains concise and professional.”

    Outcome:
    With added instructions about tone and response structure, the AI will likely generate a more helpful and customer-friendly response.

  2. Reflection:

    • Review the two outputs. How did adding empathy and a clear structure affect the AI’s response?
    • Was the refined output more appropriate for a customer service scenario?

Exercise 3: Generating Structured Summaries

In fields like education or business, creating structured summaries is a common task. This exercise focuses on improving AI-generated summaries.

  1. Task:
    Start with a general prompt for summarizing a document and refine it to produce a concise, structured summary.

    Initial Prompt:

    • “Summarize a research paper on climate change.”

    Steps:

    • Step 1: Analyze the AI’s initial summary for length and clarity.
    • Step 2: Refine the prompt to add structure and focus.
      Refined Prompt:
      • “Summarize the key findings of a 2022 research paper on climate change in 150 words. Highlight the paper’s conclusions on rising sea levels and global temperature increases.”

    Outcome:
    By specifying the word count and the focus of the summary, the AI should generate a more concise and relevant output.

  2. Reflection:

    • Compare the length, clarity, and relevance of the two summaries.
    • Did the refined prompt help in delivering a more focused and structured output?

Exercise 4: Creating Effective Lists and Instructions

This exercise focuses on improving the quality of AI-generated lists or step-by-step instructions.

  1. Task:
    Develop a prompt to create a step-by-step process for a common task.

    Initial Prompt:

    • “Explain how to bake a cake.”

    Steps:

    • Step 1: Review the AI’s initial response to see how well it structures the instructions.
    • Step 2: Add details to improve clarity and ensure a logical progression.
      Refined Prompt:
      • “Provide a clear, step-by-step process for baking a chocolate cake. Include details for mixing the ingredients, baking time, and cooling instructions. Ensure each step is concise and easy to follow.”

    Outcome:
    The refined prompt should yield a more structured and clear set of instructions, with each step logically following the previous one.

  2. Reflection:

    • How did the clarity and structure of the instructions improve with the refined prompt?
    • Was the final output easier to understand and follow?

Exercise 5: Collaborative Prompt Engineering

Collaboration in prompt engineering allows you to compare techniques and learn from others’ approaches. In this exercise, you’ll work with a peer to improve prompts.

  1. Task:
    Partner with someone and each craft a prompt for a specific task. Afterward, provide feedback on how the prompt could be improved.

    Example Task:

    • Create a prompt to generate a detailed product review for a new smartwatch.

    Steps:

    • Step 1: Write an initial prompt individually.
      Example:
      • “Write a product review for a smartwatch.”
    • Step 2: Swap prompts and review your partner’s prompt.
    • Step 3: Provide constructive feedback on how to improve the clarity, structure, or specificity of the prompt.
      Refined Example:
      • “Write a detailed product review for a new smartwatch, focusing on its fitness tracking features, battery life, and ease of use. Include pros and cons and rate the product out of 5 stars.”

    Outcome:
    Collaborating on prompt refinement can provide new insights and strategies for improving your own techniques.

  2. Reflection:

    • How did your partner’s feedback help refine your prompt?
    • Were there any techniques you hadn’t considered before?

Iteration: The Key to Effective Prompt Engineering

Iteration is an integral part of prompt engineering. The goal is to test, refine, and improve your prompts based on the AI’s responses. This continuous cycle allows you to discover which prompts work best for different tasks.

  1. Task:
    Take one of the prompts from the previous exercises and iterate on it at least three times.

    Steps:

    • Step 1: Craft a prompt and test the AI’s response.
    • Step 2: Refine the prompt based on the response, adding constraints or more detailed instructions.
    • Step 3: Test the refined prompt and further iterate as needed to improve the output.

    Outcome:
    By testing and refining a prompt multiple times, you will see how small adjustments lead to more controlled, relevant outputs.

  2. Reflection:

    • After multiple iterations, how did the output evolve?
    • What changes had the greatest impact on improving the AI’s response?

Conclusion

This lesson provides a set of hands-on exercises to help you apply and refine your prompt engineering skills. By working through scenarios that cover creative tasks, customer service, structured summaries, and collaboration, you gain practical experience in optimizing AI responses. The key takeaway is that prompt engineering is an iterative process—each iteration allows for improved clarity, relevance, and structure in AI-generated content.

The next step is to continue practicing with real-world applications, further honing your skills in prompt engineering across various fields. By consistently applying these techniques, you can master the art of guiding AI behavior for more effective and tailored results.

Explore the full course:

Overview: Prompt Engineering Course Overview

Lesson 1:  What is Prompt Engineering

Lesson 2: Crafting Effective Prompts

Lesson 3: Guiding AI Behavior

Lesson 4: Practical Applications

Lesson 5: Hands-on Exercises