Unit 2: Prompt Engineering for Teachers

Most teachers get poor results from AI because their prompts are too vague. In this unit you'll learn a four-part framework that turns a one-line request into a classroom-ready response, every time.

⏱ 55 minutes 📊 Beginner 🎯 4 learning objectives

Learning objectives

Write clear, specific prompts that produce classroom-ready output.
Apply the R.O.C.C. framework (Role, Objective, Context, Constraints).
Refine and iterate prompts to improve quality.
Save and reuse prompt templates.

Why most teachers get bad results

Vague prompts produce vague answers. Look at the difference:

Weak prompt	What you get
"Write a lesson on fractions."	A generic, American-style, 25-minute lesson with no examples your students will recognise.
"Make a quiz on electricity."	10 random questions, no answer key, no difficulty levels, written for whom?
"Explain photosynthesis."	A textbook paragraph in adult English with no concrete example.

The fix isn't a better AI tool. It's a better prompt.

The R.O.C.C. framework

Every prompt should include these four pieces, in roughly this order:

R — Role

Who should the AI be? Give it a job title.

"Act as a JSS2 Biology teacher in Lagos…"

O — Objective

What exactly do you want it to produce? Be concrete.

"…create a 40-minute lesson plan…"

C — Context

Who are your students? What's the setting?

"…for a class of 45 students with mixed ability, in a school with limited materials…"

C — Constraints

What rules must it follow? Length, language, format, examples.

"…use only locally available materials, simple English, 3 group activities, and end with 5 evaluation questions in Nigerian context."

💡 Putting it together

"Act as a Primary 5 Mathematics teacher in Nigeria. Create a 35-minute lesson plan on adding fractions with the same denominator. My class has 50 pupils. Include learning objectives, materials (chalk, cardboard, bottle caps), 3 activities, and 5 evaluation questions. Use Nigerian Naira and local food examples like jollof rice portions."

Iterative prompting

Treat AI like a junior assistant. The first draft is rarely the last draft. After you get a response, just keep talking to it:

"Now make the activities harder."
"Add 3 more questions, this time at Bloom's Apply level."
"Rewrite the lesson plan to fit 30 minutes instead of 40."
"Now produce the same lesson plan as a one-page student note."

Each follow-up costs you 10 seconds and saves you 15 minutes of editing.

Few-shot prompting (advanced)

If you have an example of exactly what you want, paste it in:

"Write a 5-question quiz on the topic 'Living and Non-living Things' for JSS1. Format each question exactly like this example:
1. Which of the following is a living thing? (a) Stone (b) Cassava plant (c) Cup (d) Pen"

The AI will copy your format precisely. This is the fastest way to get consistent output across many quizzes or worksheets.

Building your prompt library

Save your best prompts in one place — a Google Doc, Notion page, or even a WhatsApp note to yourself. Group them by purpose:

Lesson plan (one per subject)
Quiz / worksheet (with answer key)
Rubric
Parent letter / progress update
Scheme of work expansion
Student note / handout

Over six months, you'll build a library of 20–30 reusable prompts. Monday morning planning then takes 25 minutes instead of 3 hours.

🎬 Demo — Side-by-side comparison

Open ChatGPT. Submit the weak prompt first, then the strong one:

Weak: "Write a lesson on fractions."

Strong (R.O.C.C.): "Act as a Primary 5 Mathematics teacher in Nigeria. Create a 35-minute lesson plan on adding fractions with the same denominator. My class has 50 pupils. Include learning objectives, materials (chalk, cardboard, bottle caps), 3 activities, and 5 evaluation questions. Use Nigerian Naira and local food examples like jollof rice portions."

Compare the two outputs. The R.O.C.C. version will be 3–4× longer and immediately usable. That's the gap that costs teachers hours every week.

✍ Exercise 2.1 — Prompt makeover

Rewrite each weak prompt using the R.O.C.C. framework:

"Write a quiz on Civic Education."
"Explain electricity."
"Make a poem for children."

Deliverable: 3 rewritten prompts + the AI output for one of them, in your workbook.

Unit summary

R.O.C.C. = Role, Objective, Context, Constraints — use it on every prompt.
Iterative prompting beats starting over.
Few-shot examples force consistent formatting.
A personal prompt library compounds into real time-savings.

Prompt Engineering for Teachers