What Is Prompt Engineering? How to Actually Talk to AI (Free AI Course – Day 4)

Day 4 already. Over the last few lessons I kept things pretty basic — what AI is, how it works, LLMs, machine learning, all of that. If you skipped those, go back first. This one will land better once you've seen those.

Today's topic: how do you actually talk to AI? Weird question maybe, but stick with me. You want AI to do something for you, you have to tell it what. Obviously. That "telling it" part is what people call a prompt. You've heard the word by now, probably too many times. Type "write this," "explain that," "fix this for me" into ChatGPT or whatever — congratulations, you just wrote a prompt.

When I first started using ChatGPT I'd fire off one-liners and then get annoyed the answers felt flat. Took me way longer than I'd like to admit to realize — the tool wasn't the problem. My prompts were.

Simple rule, really. Good prompt, good answer. Lazy prompt, lazy answer. That's basically the entire concept behind prompt engineering — writing your instructions clearly enough that you actually get something worth reading back.

Textbook version if you need one: "Prompt engineering is the process of writing clear and effective instructions to AI, so that it produces more accurate and useful outputs." Reads simple on paper. Feels different once you try it yourself.

Good Prompt vs Bad Prompt — What's the Actual Difference

You know you're "supposed" to write good prompts now. Cool. But what does that mean when you're actually sitting there typing? Example time, forget theory for a second.

Kid at home, stuck on photosynthesis for school. You want AI to help. Type "explain photosynthesis," hit enter — you'll get some paragraph back. Could be too basic, could be way too technical. Either way it probably won't help much. Kid reads two lines, shrugs, closes the tab.

I actually ran this exact test with my nephew — 6th grade, hates science homework. Plain prompt first: he read maybe two sentences and gave up. Rewrote it properly, gave it context and a role, and this time he actually finished reading it. Understood it too. Kind of surprised me honestly.

Two things going on here.

First thing — give it a role. Obvious once someone points it out, but we already do this in real life without thinking. You wouldn't ask your math teacher a biology question. You'd find someone who actually knows biology. Same deal with AI — tell it what to "be" before you even ask your question.

Second thing — actually explain what you want. Not just the topic, the whole picture. How simple should it be, how long, what tone, do you want examples — spell it out.

A Few Bad-vs-Good Examples

Photosynthesis: "Explain photosynthesis" gets you the flat, generic version. "Act as a science teacher. Explain photosynthesis in very simple words, like you're teaching a 10-year-old. Short sentences, one real-life example." — this one actually gets read.

Role first, then details. That's the entire trick, every single time.

Story writing comes up constantly with school homework — half of India's kids seem to get "write a story about a robot" as an assignment at some point. Weak version: "Write a story about a robot." Better: "Act as a professional story writer. Write a short, funny story about a school boy who finds a talking robot in his classroom. Keep it easy to read for students."

Interview prep — a friend of mine used almost this exact wording before a sales interview: "You are an HR interview coach. Give me 5 common interview questions for a sales executive role, with simple sample answers that sound confident." Skip the role and just type "help me for interview" and you'll get something generic that could apply to literally any job. Not useful.

Career advice for students: "Act as a career counselor. Explain 5 career options for Indian students after class 12. Simple language, one or two lines each." Versus just typing "tell me about careers," which gets you nothing specific to act on.

Learning Excel in a week: "Act as a computer skills trainer. Give me a simple 7-day plan to learn Excel as a complete beginner." "How to learn Excel" alone won't get you a plan, just a vague overview.

Same pattern, every time. Role, then details.

One Prompt Is Rarely Enough

Something people get wrong — they think you write one prompt and get a finished, perfect answer immediately. Sometimes that happens, sure, lucky you. Most of the time though, the first answer is just okay, and that's where the real work starts.

Kind of like building a rangoli during Diwali. Most of us watched our mothers or grandmothers do this growing up — first you sketch the rough outline, that's your first prompt, just the basic shape of what you want. Then you go back and fix or add something, filling in colors, and it starts looking better already. One more small adjustment after that — finishing touches, the tiny details that make it actually look complete.

This shift changed a lot for how I use AI day to day. I used to give up after one try if it wasn't great. Now the first reply is just a rough draft. "Make it shorter." "Add an example." "Explain it simpler." Keep nudging until it's actually right.

Or maybe the answer's good but you wanted it in Hindi. Or bullet points instead of one giant paragraph. Just type "write this in Hindi" or "put this in bullet points" right underneath. Takes two seconds, adjusts instantly.

The whole point — keep refining, keep adding detail, until whatever comes out actually matches what was in your head to begin with.

That's Day 4, Wrapping Up

You've got a decent handle on writing prompts now, or at least the basics of it. This one thing genuinely changed how much I actually use AI day to day — went from opening it once in a while out of curiosity to relying on it for actual work. Try a couple of these examples today, see what happens with something you're already working on.

Drop a comment, tell me how this one landed for you. Likes, comments, shares — all of it helps more than people realize. Follow along on Instagram too, we post there fairly often. That wraps up prompts for today. Next chapter tomorrow.

What Is AI & How It Works

What is an LLM? Simple Guide for Beginners

Machine Learning & Algorithms Explained Simply