CRIN←◎ BeginnerWhat a Prompt DoesPress enter or space to select a node. You can then use the arrow keys to move the node around. Press delete to remove it and escape to cancel.
Press enter or space to select an edge. You can then press delete to remove it or escape to cancel.
What a Prompt Does
A prompt is not just a question — it is the entire instruction set that shapes the response. Prompts have hidden structure, and every word you add narrows the space of possible answers. Specificity is a superpower.
Course: Beginner.
This lesson covers 5 concepts: A Vague Prompt, Prompt Has Structure, What AI Actually Sees, A Better Prompt, The Output Transforms.
A Vague Prompt
Two words. AI will answer — but it has to guess everything: the audience, the length, the format, the depth, the style. All of that guessing is where quality gets lost.
AI responds to exactly what you give it. Vague input produces vague output. Your words are the only steering wheel — and right now it is barely touched.
Like ordering at a restaurant by saying "food please" — you will get something edible, but probably not what you were craving.
"Explain gravity" — is this for a child or a physicist? Two sentences or two pages? With equations or without? An analogy or a definition? AI will guess. It will probably guess wrong for you.
Prompt Has Structure
Every prompt has invisible structure: a system message that sets the rules, a user message with your request, and optionally prior assistant turns. The model sees all of them together, every time.
Understanding this structure gives you a second level of control — you can set the persona, rules, and format before your question even arrives.
Like a play with stage directions, dialogue, and character notes — the system message is the stage directions that run before you ever walk on stage.
ChatGPT has a hidden system prompt that runs before every conversation — shaping tone, safety behaviour, and response style. You never see it. But it shapes everything you read.
What AI Actually Sees
Your two-word prompt has been placed into structure — a system message above it, your message below. AI sees both together. But "You are a helpful assistant" is doing almost nothing to guide the response.
The system message is the most powerful part of a prompt. Here it says almost nothing specific — which is exactly why vague prompts produce generic answers.
What you typed as two words gets wrapped in structure — like putting a note into an official form before it reaches the desk. The form adds context. Right now, the form is almost blank.
System: "You are a helpful assistant." That tells AI almost nothing useful. User: "Explain gravity." Two words. Together they leave AI with a universe of equally valid responses.
A Better Prompt
Same topic, completely redesigned. The system message sets who AI is and how it works. The user message says exactly what to produce and how. AI now has almost no room to guess.
The model did not get smarter. The instructions got clearer. That clarity alone is enough to transform the output completely.
The difference between a vague request and a clear one is the difference between getting something and getting exactly what you wanted.
System defines persona + style + rules. User defines topic + format + analogy + length. Together: a brief so precise the response almost writes itself.
The Output Transforms
The specific prompt produced a specific, useful answer — exactly the format requested, with the analogy requested, for the audience specified. The prompt wrote the brief. The model executed it.
Prompting is the skill layer on top of AI. The model stays the same. A well-written prompt can make it perform dramatically better on your specific task — and that skill is entirely yours to develop.
AI is like a powerful tool — a better grip gets you much further. The prompt is the grip. Sharpen it, and everything that follows gets sharper too.
"• Gravity is like Lego studs — every brick wants to snap onto every other." — Concise, memorable, age-appropriate, exactly on format. The prompt made all of this possible.