CS: Prompt Engineering and Working with LLMs
Writing clear prompts, checking outputs, and using AI responsibly
CS: Prompt Engineering and Working with LLMs
Writing clear prompts, checking outputs, and using AI responsibly
CS - Grade 6-8
- 1
A student types this prompt into an LLM: "Tell me about dogs." Rewrite the prompt so it is more specific and useful for a 7th grade science report about how dogs communicate.
Include the audience, topic, format, and details you want.
A stronger prompt would be: "Write a 7th grade science explanation of how dogs communicate using body language, sounds, and scent. Include three examples and keep the response under 250 words." This prompt is better because it gives the topic, audience, length, and details to include. - 2
List three parts of a strong prompt for an LLM. For each part, explain why it helps the model give a better answer.
A strong prompt can include a clear task, context, and output format. The clear task tells the model what to do. The context gives background information so the answer fits the situation. The output format tells the model how to organize the response, such as a list, paragraph, or table. - 3
Look at this prompt: "Make it better." Explain why this prompt is weak, then write a stronger version for improving a paragraph in a history essay.
A good prompt tells the model what text to work on and what kind of changes to make.
The prompt is weak because it does not say what "it" is or what kind of improvement is needed. A stronger version would be: "Revise this history essay paragraph to make the main idea clearer, improve word choice, and keep the meaning the same. Then list the changes you made." - 4
An LLM gives this answer: "The capital of Australia is Sydney." What should you do before using this answer in a school assignment, and what is the correct information?
I should check the answer with a reliable source, such as an encyclopedia, official government website, or trusted atlas. The correct capital of Australia is Canberra, not Sydney. - 5
Write a prompt that asks an LLM to help you study for a math quiz on ratios. The prompt should ask for practice questions, step-by-step explanations, and feedback after each answer.
Ask the model to take a role, give one task at a time, and respond after you answer.
A good prompt would be: "Act as a helpful math tutor for a 7th grade student studying ratios. Give me one practice question at a time, wait for my answer, then explain the solution step by step and give feedback before moving to the next question." - 6
The diagram shows a prompt improvement cycle: Ask, Check, Revise, Try Again. Explain how this cycle can help someone get better results from an LLM.
The cycle helps because the first prompt may not give the best answer. After asking, the user checks whether the response is accurate and useful. Then the user revises the prompt by adding details or correcting confusion. Trying again can lead to a clearer and more helpful response. - 7
A student wants help writing a book report. Which prompt is more responsible, and why? Prompt A: "Write my whole book report so I can turn it in." Prompt B: "Help me make an outline for my book report and ask me questions about the book so I can write it myself."
Think about the difference between getting help and having someone else do the work.
Prompt B is more responsible because it helps the student learn and create their own work. Prompt A asks the LLM to do the assignment for the student, which could be dishonest if turned in as original work. - 8
An LLM can sometimes make up facts, sources, or details. What is this problem often called, and what are two ways a student can reduce the risk of using false information?
This problem is often called a hallucination. A student can reduce the risk by checking facts with reliable sources and asking the LLM to show how it knows something, while still verifying the information independently. - 9
The chart compares two prompts. Prompt 1 says: "Explain climate change." Prompt 2 says: "Explain climate change in 5 bullet points for a 6th grader, using one example of how it affects oceans." Identify two reasons Prompt 2 is likely to get a more useful answer.
Look for details about audience, format, length, and content.
Prompt 2 is likely to get a more useful answer because it gives a specific format, which is 5 bullet points, and it names the audience, which is a 6th grader. It also asks for a specific example about oceans, so the response will be more focused. - 10
Write a short safety rule for using LLMs at school. Your rule should mention personal information and checking important answers.
A good safety rule is: "Do not share personal information with an LLM, such as your full name, address, passwords, or school ID, and always check important answers with a trusted source or teacher." This rule protects privacy and helps prevent mistakes.