Prompt engineering

Prompt engineering

Objective: 

Learners will acquire the ability to formulate effective and precise prompts for AI systems (such as ChatGPT or Gemini) using a selected prompting method in order to achieve customized and structured results.

Content and methods: 

The worksheet introduces a method of prompt engineering. It explains the logic behind this method and provides a detailed example prompt. The method is illustrated as a “blueprint” for the perfect prompt. Learners should apply the method to their own use cases and create a flashcard.

Skills:

  • Methodological competence in dealing with AI systems
  • Ability to structure complex tasks into logical, sequential steps

Target group:

Grade 7 and above

PY
RB
SE
TH

55 other teachers use this template

Target group and level

Grade 7 and above

Subjects

non-subject specific content

Prompt engineering

Icon

Introduction

You've probably used ChatGPT, Gemini, or another AI program a few times. You may have noticed that the answers are sometimes very good, but sometimes less helpful. This is usually due to the prompt.

A prompt is the input or instruction you give to an AI. It determines what the AI should do and how exactly it responds.

Prompt engineering is used to ensure that AI systems provide the best and most appropriate responses possible. This involves the targeted formulation of prompts in order to obtain better, more accurate, or more creative results.

Depending on your purpose (e.g., gathering information, writing texts, or developing ideas), different methods of prompt engineering are suitable.

In this worksheet, you will find instructions for a method that will teach you how to formulate effective prompts.

The Blueprint Method: Precision Prompting with Instruction Tuning

Let's be real: have you ever asked an AI to help you with a school project, maybe to create a list of pros and cons, and it gave you back a huge, messy paragraph instead? The problem isn't that the AI doesn't know the information; it's that it doesn't understand exactly how you want that information presented. This is where a powerful prompt engineering technique called "Instruction Tuning" comes into play.

Think of it like you're building a complex model kit. You wouldn't just look at the picture on the box and start guessing. You'd want a detailed, step-by-step instruction manual. With Instruction Tuning, you stop being someone who just asks a question and become the architect of the AI's answer, providing it with a precise blueprint to follow.

The Core Idea: Moving from "What" to "How"

While other methods might give the AI a role or some context, Instruction Tuning focuses on the process itself. You break down your big request into a series of smaller, logical commands. Instead of just telling the AI, "Analyze this historical event," you guide it like a mentor, giving it a chain of commands to execute in a specific order.

The incredible advantage of this is control. You leave no room for the AI to guess how the final output should be structured. This is a game-changer for schoolwork, where the format of your answer is often just as important as the content. You force the AI to think and work according to your rules, leading to incredibly accurate and tailor-made results.

Your Blueprint for the Perfect Instructional Prompt

Creating a strong prompt with this method is all about organizing your thoughts into a crystal-clear set of directions.

First, you define the overall goal. What is the final product you want to receive? A character analysis for your English class? A study guide for a biology test? This sets the overall frame for the task.

Next comes the core of the method: the step-by-step instructions. This is where you break the main task into smaller, digestible commands. Numbering your steps is key, as it gives the AI a clear sequence to follow. Each step should describe a single, well-defined action. A common pitfall here is being too vague. An instruction like "make it short" is hard for an AI to interpret. It's much better to be specific: "Write a summary that is a maximum of 60 words."

Finally, you establish the constraints and rules. This section defines the format, style, or any limitations. Should the AI use bullet points? Should it adopt a formal and objective tone? These rules act like guardrails, keeping the AI on track and preventing it from veering off course.

Instruction Tuning in Action: A School-Ready Example

Imagine you need to analyze a famous poem for English class and need a structured breakdown to help you write your essay.

Here’s how you can combine the building blocks into a powerful prompt:

Example Prompt:

[Goal]
Create a structured analysis of William Shakespeare's "Sonnet 18" ("Shall I compare thee to a summer's day?"). The goal is to produce a clear, foundational guide for a high school essay.

[Instructions]
Please follow these steps in this exact order:

  1. Write a three-sentence summary of the poem's main message.
  2. Identify and state the central theme of the poem in a single sentence.
  3. List two distinct literary devices used in the poem. For each device, provide the specific quote from the poem and a one-sentence explanation of its effect.
  4. Conclude with a single sentence that explains the poem's lasting significance.

[Format & Rules]
Style: Use a formal, academic tone.
Format: Structure the entire response using the following headings: "Summary," "Theme," "Literary Devices," and "Conclusion." Use a numbered list for the literary devices.
Rule: Do not use any first-person pronouns (I, me, my).

With a prompt like this, you're not leaving anything to chance. You can use this method for almost anything, from creating outlines for presentations to generating step-by-step solutions for science problems. Go ahead, try it out and see how you can turn your AI into a precise and personal assistant.

📝 Think of three examples from your everyday life or teaching where it would be useful to use this prompt method.

📌 Create a cheat sheet for this prompt method so you can see at a glance what you need to pay attention to when prompting. Fill out the table to do this.

Name
Possible uses/purpose
Possible sources of error

📝Enter the individual steps for prompting and a corresponding example.

✅ Example index card

Name
Possible uses/purpose
Possible sources of error

📝Enter the individual steps for prompting and a corresponding example.