“This makes a huge difference! I wish I had known this months ago!” 

Teachers across Kent County are beginning to experiment with how artificial intelligence can support learning in their classrooms. We often hear comments like this when we introduce teachers to prompt engineering. Prompt engineering skills are often what differentiate uninspired AI content from practical, useful learning materials. Better prompts help us differentiate with more precision, making it easier to ensure every student has an entry point into the learning.

In other words, basic prompts get basic results, while well-crafted prompts get far better results. Consider this image of a kangaroo on a beach created with a basic prompt. Compare it to the following image created with a much more detailed and specific prompt. Prompt Engineering – crafting a detailed and specific prompt to generate higher-quality materials with AI. 

Created with ChatGPT. Prompt: A kangaroo on a beach.

Created with Gemini: Create a photorealistic image of a kangaroo sunbathing on a sandy beach. The kangaroo has beautifully manicured purple fur, long wavy hair atop its head, wears large aviator sunglasses, and has a very, very chill smile, as if it is quietly having the best time in its life. The kangaroo is sipping a tropical drink in a coconut shell through a straw. The sandy beach sits on an idyllic blue coastline, lined with palm trees and other sunbathers in the background. 

Components of a High-Quality Prompt

The basic idea of prompt engineering is to be very specific about what you want, so you get close to what you want from your AI tools. There are many prompting frameworks developed over the last few years, but they all boil down to about six components. 

All AI systems generate better outputs when they…

1. Have a GOAL: Be very clear and specific about the exact task you want the AI tool to complete.
2. Have CONTEXT: Provide background information such as grade level, standards, curriculum, and audience to set the scene for the AI tool.
3. Have a ROLE: Give the AI tool an identity, background, or expertise, so it knows what domains are relevant to your request.
4. Know what FORMAT you want the output to be in: Tell the AI tool exactly what you want the output to include. Consider things like bulleted lists, rubrics, tables, or file formats. Provide example documents if available.
5. Know what CONSTRAINTS they need to work within: Define the nuance your audience needs. Consider factors such as a 3rd-grade reading level, a professional tone, or parent-friendly language.
6. Have a HUMAN using their judgment to verify the output: All AI tools have the potential for bias and fabrication. It’s important that humans apply their ideas, oversight, and judgment to their work with AI tools.

You can find these six elements in most prompting frameworks. In this post, we will explore the PREPARE framework from The AI Classroom. There are many more! Links to other frameworks that you can explore later:

• PREPARE – Fitzpatrick, Fox, Weinstein
• REFINE – Catlin Tucker
• SPARK – Sabba Quidwai
• RISEN – Kyle Balmer
• CO-STAR – Sheila Teo

The PREPARE Framework

The framework we use most often is the PREPARE framework, from The AI Classroom. We find it to be well-rounded, easy to remember, and helpful for educators. It looks like this:

Prompt – Introduce the task

Role – Give the AI a role or focus

Explicit – Provide instructions and resources for context

Parameters – Set the structure of the response

Ask – Ask AI to ask you follow-up questions

Pro-tip: AI tools make mistakes when we force them to make assumptions. Asking AI to follow up with questions usually uncovers your assumptions and provides clarity.

Rate – Ask AI to rate its response and suggest improvement areas

Emotion – Set the tone of the response

From Theory to Practice: The ‘Uberprompt’ in Action

Let’s imagine you’re a 9th-grade US History teacher and you want to create a choice board to deepen your students’ understanding of the US’s involvement in World War II. A simple prompt might be 

Create a choice board for students to learn about the US involvement in World War 2.

A PREPARE prompt for the same situation might look something like this: 

PREPARE Framework Element	Prompt Text Example
PROMPT: Introduce the task	Create a choice board to help students deepen their understanding of the US involvement in World War 2.
ROLE: Give the AI a role or focus	You are a 9th-grade US History teacher. You are an expert at incorporating engagement techniques.
EXPLICITProvide instructions and resources for context	The choice board will address content in the attached lesson plan.All choices on the choice board should focus on at least one high school C3 social studies skill from page 86 of this document.
PARAMETERS: Set the structure of the response	The choice board should have nine options for students, each aligned with a different learning style. Each option should have an exemplar
ASK: Ask AI to ask you questions	Ask me questions about anything you are uncertain about before you begin to build your response.
RATE: Ask AI to rate its response	Please rate your response on how well it meets all of the criteria I’ve given you. Suggest areas that may be improved.
EMOTIONS: Set the tone of the response	The tone of the language on the choice board should be engaging and student-friendly for 9th-grade students.

Try copying these two prompts into Gemini or your preferred AI tool to see what you get. Do you notice a difference? Which do you think has the most potential to enhance student learning?

The PREPARE framework can feel like a lot to think through, especially at first, but taking the time to think through these components can save you a LOT of time and frustration. As an added bonus, well-crafted prompts typically require less re-prompting. Less re-prompting generally means our AI tools use less energy, reducing carbon emissions and water consumption. 

After using the framework a few times, PREPARE will become a natural part of your prompting strategy, allowing you to adjust pieces based on what’s relevant. After perfecting your prompt, you may end up with a highly specific, repeatable prompt. We’ve been calling this kind of prompt an uberprompt. If you’ve gotten to this point, you might consider creating a Gem or custom GPT for your prompt so you don’t have to type in the prompt every time you chat.

The Human in the Loop – Mitigating Bias and Fabrication

Even with perfect prompts, all AI systems have biases and can fabricate misinformation. If we don’t use our human and professional judgment, we can unintentionally spread misinformation or perpetuate bias, causing harm and impeding learning. It’s crucial that we take a moment after an AI system generates material for us to analyze the output. We often teach Sabba Quidwai’s 10/70/20 rule to emphasize this point.

The 10% human spark is the creativity, ideas, and knowledge we provide to our AI tool. Then our AI tool does about 70% of the work, generating the materials for us. It is then our job to use our human judgment (20%) to review and modify the material so it works for our learners, our context, and our curricular goals. Not only does this ensure that your students receive more accurate, less biased learning, but it also encourages melioration (learning through iteration) while reducing cognitive atrophy (letting AI do the thinking for us). 

Explore More

Prompt engineering and melioration are skills we develop through practice. If you’re ready to start building your own uberprompts, we have several tools to help you along the way. You can utilize our PREPARE template to help you think through the framework components, or browse our list of Prompting Libraries to save time with pre-made examples for educators. If you want to dive deeper, feel free to use or remix our full training materials for your own building.

Don’t forget to join us for Promptober this October! We’ll be sharing daily tips and tricks to help you keep sharpening your skills. If you’d like some hands-on help or want to talk through a specific AI project for your classroom, reach out to our EdTech team – we’d love to support you and your students!

This post was written by Keith Tramper, Education Technology Consultant for Kent ISD, and edited by Amanda Walma, Teaching and Learning Marketing and Communications Specialist for Kent ISD. In collaboration with our team, artificial intelligence was utilized to assist in the editing process and the creation of all images in this post.

Walk the Walk: How We Made Our Featured Image

We don’t just teach prompt engineering; we use it. To practice the ‘Human Spark’ and ‘Human Judgment’ we discussed, we avoided a stock photo and instead used our own PREPARE framework to generate this featured image:

---

The Uberprompt We Used:

“A photorealistic, wide-angle landscape photograph captured by a professional photographer for Kent ISD. A female middle school teacher is leading a small, diverse group of students in an engaging lesson. She is smiling genuinely at one of the students, representing the ‘Human Spark.’ She is standing next to a large digital screen, which is displaying a vibrant, customized ‘World War II Choice Board’ table. On the table, some cells feature small icons representing ‘Propaganda Poster’ and ‘Battle Map,’ which are clearly AI-generated elements. Students are actively collaborating in pairs with laptops, and some are interacting with the digital board. The classroom is modern, bright, and with natural light and colorful learning materials. The focus is on the teacher and the interaction, with the screen clearly visible and data nodes subtly connecting to the laptops. The background is a slightly blurred but detailed classroom environment. 680×453 pixels. Shot on a Canon EOS-1D X Mark III.”