The Prompt Engineering Playbook (for Educators)
The ultimate guide to writing expert prompts for AI.
Prompt engineering refers to designing high-quality prompts that guide machine learning models to produce accurate outputs. In other words, it is the process of crafting tasks, statements, questions, or scenarios as inputs for an AI tool to respond to and involves:
Selecting the correct type of prompt;
Optimize the length and structure of the prompt, and;
Determining their order and relevance to the task.
For example, consider the following prompt that a teacher might enter into ChatGPT:
Create a lesson plan that helps teach the concept of self-management to my students.
The output:
You can see how the generative AI’s response is a decent attempt at putting together a lesson plan… but it is quite long and does not include the type of step-by-step instructions or facilitation guidelines that a teacher might expect.
That’s because when it comes to effectively using generative AI tools such as ChatGPT, the process of writing (and refining) is incredibly important. Without the use of advanced prompting techniques, responses from AI tools can be very poor and unhelpful.
Instead, the goal is to provide clear and specific instructions to an AI model on the type of output we desire. This ensures that the content it generates is clear, accurate, relevant, and meets our needs.
This guide is designed to help you learn about the research behind prompt engineering and apply it to craft effective prompts. Continue reading to review an overview of prompt engineering fundamentals for educators and dive into some best practices for writing expert prompts.
—
Prompt Engineering 101 for Educators
Generative AI is a type of artificial intelligence that can generate new content similar to the examples it has been trained on. Generative AI learns patterns and relationships from the data you input and uses it to generate new, unique outputs.
Generative AI tools (such as ChatGPT) require a prompt that is typically in the form of text, although there are other inputs that the AI systems can interpret. The AI then uses an algorithm to produce new content in response to the prompt. The resulting content might provide an answer to a question, a summary of information, instructions to follow, or numerous other outputs as directed by the original prompts.
The below visual from Educative’s course on prompt engineering gives an overview of this process.
When using generative AI tools, a prompt is the input provided to the model in order to generate a response. The prompt can be a question, a sentence, a paragraph, or a set of instructions.
Types of Prompts
Prompts may be utilized for accomplishing a variety of tasks. Grasping how to compose prompts to effectively reach the intended outcome is crucial. We'll explore diverse examples of prompts and how to employ them in various situations.
Text Summarization (e.g., generating a summary of a longer text)
Information Extraction (e.g., identify and extract relevant information from a given document or dataset)
Questions and Answers (e.g., generating responses to questions)
Text Classification (e.g., classify or categorize a text into predefined classes; for example, read a customer review and classify it as positive, negative, or neutral)
Translation (e.g., generate text translations from one language to another)
Code Generation (e.g., create or complete code)
Reasoning (e.g., apply logical reasoning and draw conclusions based on provided information)
—
Guiding Principles for Writing Prompts
(1) Always tell the AI who it is and what you want it to do.
Context aids the AI in generating customized responses that are beneficial. This can be accomplished through role prompting, a technique in prompt engineering to control the output generated by assigning the model a specific role. We can ask ChatGPT, for instance, to be a curriculum designer, a researcher, a poet, a historical figure, etc.
Effective role prompting occurs when we give the AI tool some context about how we expect it to respond given its particular role or persona. We can set the scene for the AI by sharing what role, expertise, and/or environment it should use to guide its output. The more information and background details we provide in our prompt, the better our chances of generating an accurate and relevant response.
Here are few examples of role prompting in practice:
Early Literacy Curriculum Designer
You are an experienced educator and curriculum designer with expertise in writing and early literacy for elementary-aged students. Author an in-class activity for students to practice "phoneme-grapheme mapping." When writing the instructions, make sure to use age-appropriate language, specify learning objectives, and use asset-based language.
Diagnostic Quiz Creator
You are a quiz creator of highly diagnostic quizzes. You will make good low-stakes tests and diagnostics. You will construct several multiple choice questions to quiz my audience on the topic. The questions should be highly relevant and go beyond just facts. Multiple choice questions should include plausible, competitive alternate responses and should not include an "all of the above option." At the end of the quiz, you will provide an answer key and explain the right answer.
Coach
You are a supportive, friendly, and helpful coach assisting a student in reflecting on a recent team experience. Your role is to guide them through a thoughtful exploration of their challenges, insights, and growth. Follow the conversation guide below, adapting your questions based on the student's responses.
Keep reading with a 7-day free trial
Subscribe to Empathy in Bytes to keep reading this post and get 7 days of free access to the full post archives.