Role: who the AI should be. Task: the one job, in verbs. Context: the facts it cannot guess. Format: exactly how the output should look. Constraints: length, tone, what to avoid. Five lines, dramatically better output.
Weak vs strong, side by side
Weak: 'Write about climate change.' Strong: 'You are a science teacher for Grade 9 learners in Eswatini. Explain three causes of climate change in 150 words, using one local example. Output as three bullet points, simple English, no jargon.' Same topic, completely different result.
One task per prompt
Models degrade when a prompt smuggles in five jobs. Split multi-job requests into a chain of prompts, feeding each answer into the next. Sequencing is a Module 2 superpower.
Key terms
Prompt
The instruction text you give a model.
Role prompting
Assigning the model a persona/expertise.
Constraints
Explicit limits: length, tone, exclusions.
Chain prompting
Splitting one big job into ordered prompts.
🛠 Hands-on task: Prompt surgery
Tools: Any free chatbot
Take 3 weak prompts you have used before.
Rewrite each with Role, Task, Context, Format, Constraints.
Run old and new versions; screenshot the difference for your PoE.
Key takeaways: Before you go, three takeaways from this lesson. One: Name the five parts of a strong prompt. Two: Rewrite a weak prompt into a structured one. Three: Explain why vague prompts produce vague answers. And your hands-on task: prompt surgery. Take 3 weak prompts you have used before. When it is done, mark the lesson complete on your dashboard, keep the artifact for your portfolio, and step into the next lesson. I will see you there.
