Prompts & Prompt Engineering

KEY TAKEAWAY

A prompt is your instruction to an AI — and how you write it dramatically changes what you get back.

💡 EVERYDAY ANALOGY — BEFORE THE TECH

🧑‍🍳

Prompting is like giving instructions to a brilliant but very literal chef

If you say "make me food", you might get toast. If you say "make me a 500-calorie vegetarian lunch that takes under 20 minutes and tastes good cold", you get exactly what you wanted. The chef is equally skilled in both cases — you just gave better instructions. AI is the same. Better prompts = better answers.

👁️ Visual Explanation

Prompt anatomy: Role + Context + Task + Format + Constraints = Great output

🌍 In Plain English

AI answers are only as good as the questions you ask. A vague prompt gives a vague answer. A specific, well-structured prompt gives a useful, precise answer. Prompt engineering is the skill of writing effective AI instructions.

Real-World Example

"Summarise this article" vs "Summarise this article in 3 bullet points for a non-technical CEO who needs to make a decision in 5 minutes." Same article. Completely different usefulness.

🎓 For Students

"Explain photosynthesis" gives a textbook answer. "Explain photosynthesis using a food factory analogy, suitable for a 16-year-old studying for an exam" gives something actually useful for studying.

💼 For Professionals

A marketer who learns prompting can get AI to write in their exact brand voice, at the right length, for the right channel — reducing revision cycles from 5 to 1.

💬 Example Prompt to Try

"You are a friendly career coach. I am a 22-year-old marketing graduate looking for my first job in 2025. Give me 5 concrete tips for my LinkedIn profile, each with one specific action I can take today."

🔧 Technical DepthOptional — expand when ready

LLMs are sensitive to how instructions are framed. Techniques include: zero-shot (just ask), few-shot (show examples first), chain-of-thought (ask it to think step by step), role prompting (assign a persona), and system prompts (persistent instructions that set behaviour for the whole conversation).

Key Terms

Zero-shot promptAsking the AI to do something without giving any examples

Few-shot promptGiving 2-3 examples before asking the AI to continue the pattern

Chain-of-thoughtAsking the AI to "think step by step" before answering, which improves accuracy on complex tasks

System promptBackground instructions that shape how the AI behaves throughout a conversation

⚠️ Common Misconception

Prompting isn't magic. It's clear communication. The same communication skills that make a good email or brief also make a good prompt.