Complete AI Mastery Course

{"contentFormat":"slides.v1","completion":{"requireAllSlides":true,"requireQuiz":true},"slides":[{"kind":"title","eyebrow":"Module 2: Practical Prompt Engineering","title":"The Art of Prompting: Structure and Clarity","body":"Talking to an AI like GPT-4o, Claude, or Gemini is a skill. The difference between a frustrating answer and a brilliant one often comes down to how you frame the request. This lesson teaches you the single most valuable habit: writing prompts that are structured and clear. Whether you're drafting a business proposal in Lagos, a lesson plan in Manila, or customer emails in São Paulo, this foundation will serve you in every AI interaction.","outcomes":["Identify the five building blocks of a strong prompt: role, task, context, format, constraints","Structure prompts using labels, headings, and delimiters","Write clear, specific instructions that remove ambiguity","Apply positive framing and front-load important instructions","Rebuild a vague prompt into a well-organised one that produces reliable output"],"narration":"Welcome to the art of prompting. In this lesson, you'll learn how to structure your requests so AI models understand exactly what you need. We'll cover the five building blocks, techniques for clarity, and real-world examples from around the globe. By the end, you'll be able to turn a vague one-liner into a precise, effective prompt."},{"kind":"content","heading":"The Five Building Blocks of a Strong Prompt","body":"Think of a good prompt like a work brief for a new colleague. The best prompts usually contain some combination of these parts:\n\n- Role — who the AI should act as. \"You are an experienced accountant.\"\n- Task — the specific action to perform. \"Explain the difference between cash-flow and profit.\"\n- Context — background the model needs. \"The reader runs a small retail shop and has no finance training.\"\n- Format — how the answer should be shaped. \"Answer in 3 short bullet points.\"\n- Constraints — boundaries and rules. \"Use simple English. Avoid jargon. Keep it under 100 words.\"\n\nYou won't always need all five, but naming them mentally stops you from leaving out the piece the model needs most.","callout":{"variant":"tip","title":"Pro Tip","text":"Start every prompt by mentally checking: Role? Task? Context? Format? Constraints? Even including just 3 of these dramatically improves output quality."},"narration":"Let's break down the five building blocks. Role tells the AI who it is. Task is the action. Context gives background. Format shapes the answer. Constraints set boundaries. Using these consistently will make your prompts much more effective."},{"kind":"content","heading":"Structure: Give the Model a Skeleton","body":"Structure means visibly organising your prompt so each part is easy to identify. Long, run-on requests confuse both people and machines. Two reliable techniques:\n\n- Use labels or headings. Write Task:, Context:, Format: on separate lines. Models are trained on huge amounts of structured text and respond well to it.\n- Use delimiters — clear separators like triple quotes \"\"\", XML-style tags <text>...</text>, or dashes — to mark where your instructions end and your data begins. This prevents the model from confusing the thing you want it to work on with the instruction itself.\n\nAnalogy: Imagine ordering food at a busy counter. \"Something with chicken\" invites a random result. \"One grilled chicken wrap, no onions, mild sauce, to take away\" leaves nothing to chance. A structured prompt is the second order.","callout":{"variant":"insight","title":"Why This Works","text":"LLMs (Large Language Models) like GPT-4o, Claude, and Gemini are trained on vast amounts of structured text — code, tables, formatted documents. They naturally perform better when your prompt mirrors that structure."},"narration":"Structure helps the AI parse your request. Use labels like 'Task:' and 'Context:' on separate lines. Use delimiters like triple quotes to separate instructions from data. This makes your prompt clear and easy to follow."},{"kind":"content","heading":"Clarity: Remove Every Guess","body":"Clarity means each instruction has only one reasonable interpretation.\n\n- Replace vague words with specifics. \"Make it short\" is subjective; \"Keep it under 80 words\" is measurable.\n- State the audience and purpose. \"Explain blockchain\" differs hugely from \"Explain blockchain to a 60-year-old shopkeeper who has never used it.\"\n- Say what you do want, not only what you don't. \"Don't be too technical\" is weaker than \"Use everyday words and one real-life example.\"\n- Front-load the most important instruction. Put the core task early; models weight early instructions heavily.\n\nPositive framing: Models follow concrete, positive instructions more reliably than negative ones. Instead of \"Don't write a long boring intro,\" try \"Open with one punchy sentence that states the main benefit.\"","callout":{"variant":"warning","title":"Common Mistake","text":"Avoid double negatives or vague prohibitions. Saying 'Don't be too technical' leaves the model guessing. Instead, specify the level: 'Use language a high school student can understand.'"},"narration":"Clarity removes ambiguity. Be specific about length, audience, and tone. Use positive instructions. Front-load the most important part. This ensures the AI doesn't misinterpret your request."},{"kind":"content","heading":"Real-World Examples","body":"1. Small business owner, Nairobi (Kenya). Amina sells handmade soap and needs product descriptions. Her first attempt — \"write a description for my soap\" — returns generic filler. Restructured: \"You are a copywriter. Task: write a 40-word product description for a handmade lavender soap. Audience: eco-conscious online shoppers. Tone: warm and natural. Format: one short paragraph, end with a gentle call to buy.\" The second version is instantly usable on her store.\n\n2. University tutor, Kraków (Poland). Marek uses Claude to build quiz questions. Vague prompts give inconsistent difficulty. He adds structure: Role: physics examiner. Task: write 5 multiple-choice questions on Newton's laws. Level: first-year undergraduate. Format: question, 4 options labelled A–D, then the correct letter on a new line. Now every batch is consistent and exam-ready.\n\n3. NGO coordinator, Jakarta (Indonesia). Sri drafts community health messages in plain Bahasa-influenced English for volunteers. She specifies: \"Rewrite this notice at a primary-school reading level, under 60 words, friendly tone, keep the vaccination date and venue exactly as written.\" The clarity constraints protect the critical facts while simplifying the language.","callout":{"variant":"exercise","title":"Try It Yourself","text":"Pick one of these scenarios and write a structured prompt for it. Use the five building blocks. Then test it with a free AI chatbot (ChatGPT, Claude, or Gemini all have free tiers)."},"narration":"Let's see structure and clarity in action. Amina in Nairobi improved her product descriptions by specifying role, task, audience, tone, and format. Marek in Kraków got consistent quiz questions by adding role, task, level, and format. Sri in Jakarta protected critical facts by setting clear constraints. These examples show how a few tweaks transform output."},{"kind":"quiz","heading":"Check Your Understanding","questions":[{"question":"Which of the following is NOT one of the five building blocks of a strong prompt?","options":["Role","Task","Tone","Context"],"questionId":"cmrf73k3q000fpd27xg3kywjj"},{"question":"Why is it effective to use delimiters like triple quotes in a prompt?","options":["They make the prompt look more professional.","They clearly separate instructions from the data the model should work on.","They are required by all AI models.","They increase the word count of the prompt."],"questionId":"cmrf73k3q000gpd27255fppnu"},{"question":"What is the best way to replace the vague instruction 'Make it short'?","options":["Keep it brief.","Don't make it long.","Keep it under 80 words.","Make it concise."],"questionId":"cmrf73k3q000hpd27vsm3z21y"}],"narration":"Now let's test your understanding with a quick quiz. Three questions to check you've grasped the key concepts.","quizId":"qz_cmk7lff7l0003g4p8ihfzv75y"},{"kind":"summary","heading":"Key Takeaways","takeaways":["Use the five building blocks: Role, Task, Context, Format, Constraints.","Structure your prompt with labels and delimiters for clarity.","Be specific: replace vague words with measurable constraints.","Front-load the most important instruction.","Use positive framing instead of negative prohibitions.","Test and iterate: compare lazy vs. structured prompts to see the difference."],"narration":"To wrap up: remember the five building blocks, structure your prompts with labels and delimiters, be specific, front-load key instructions, and use positive framing. Practice by rewriting a vague prompt into a structured one. This skill will serve you in every AI interaction."}]}