Prompt Like a Pro
AI gives better answers when it gets clear, specific questions.
This section shows you how to:
- Write stronger prompts
- Fix vague or weak ones
- Build deeper prompts that go beyond just getting the answer
Why crafting prompts matters
Generative AI can explain proofs, create practice sets, or debug student reasoning—but only if the request tells it exactly what you need. A well‑built prompt:
- narrows the topic and the audience ("Explain the distributive property to 8ᵗʰ‑graders").
- states the desired output ("essential bullet points + numerical example").
- clarifies the teaching goal ("highlight why the rule works, not just how").
The payoff is tighter math language, fewer hallucinations, and responses you can drop straight into a lesson plan.
Here is a quick overview video highlighting the basics of a good prompt when working with AI chatbots.
Keep reading below for more information.
Start Simple
Get the Basics Right
Bad prompt:
"Help me with this."
Better prompt:
"Can you solve 3x + 5 = 20 and explain the steps?"
Best prompt:
"Act as a math tutor to help me understand how to solve 3x + 5 = 20. Explain it using simple, clear steps. Point out any common errors."
Be clear about what you're asking. Include the math problem and the kind of help you want (solve, explain, check, etc.).
For additional personalization, add details about how you want the response formatted, and what details the response should include.
Chain Prompts
Often, you aren't exactly sure what sort of responses are possible, or what you would like the end product to look like. Alternately, you may have a very specific idea in your head, but that idea shifts or changes as another good idea pops into your head. Humans are astonishingly creative. Ideas can spark new ideas, and these can all snowball until a final product is created. This is often the case when creating using generative AI. Thankfully, the format of these generative AI tools allows for a sort of conversation to take place. Chain prompts consist of a back-and-forth "conversation" between yourself and the computer in which you respond to the responses, and it responds to yours. Chain prompts are useful for refining and polishing. They can also be used to explore possibilities.
1. Start your prompt with a description of who you are, or who you want the computer to pretend to be. Think of this as an introductory handshake with your own personal assistant or tutor. This also provides a target audience for the design.
I am a 6th grade student
2. Next, describe what you want to learn.
I want to learn how to add fractions with unlike denominators
3. Now, describe how you want to learn.
Can you give me a brief outline?
4. Revise! This is the most important step.
Take a look at the results your prompt gives and ask for ANY changes you would like. Was the reply too long? Too short? Too confusing? Again, think of the AI chat as an assistant or tutor. Ask for any revision you can think of. The more specific your feedback, the better the results.
Make the outline funny and highlight the really important information, also include a saying to help me remember the information.
5. Always check the response. Like humans, AI is NOT INFALLIBLE. There are a few ways you can do this, depending on your confidence in the response. The first thing you can do is ask the AI to tell you where it gets its information. You can also ask it to give you websites to check the work. You can also put the response into another AI filter and ask it to check the work. With math equations, an online solution tool like Wolfram Alpha is a great way to check the work.
Here are possible results from the prompts provided above.
* I would start with this and then make changes based on my audience. For example, most of my students would find some of this so bizarre that learning would be sideswiped - unless I had the right energy to sell it. Sometimes you can make oddness work for you because of the sheer novelty.
Courtesy of "Math explanation powered by ChatGPT (OpenAI), customized for student learning."
