Lesson 1: What a Prompt Is, and Why It Matters

Learning goals:

• Understand what a prompt is and what it does
• Tell a vague prompt apart from a clear one
• See how prompt quality shapes what the AI gives back

Prerequisites: You've used an AI chat tool before | Next: Lesson 2 >>

You're Already Using Prompts (Maybe Not Well)

You open ChatGPT or Claude, type "write me a report," and get back something generic that doesn't line up with what you actually needed. That's not the AI failing you. Your prompt was too vague. All the AI saw was four words. It has no idea who the report is for, what problem it should solve, what tone to strike, or what to put in it.

A prompt is the instruction you give the AI¹. It sets how much of your intent the AI can pick up, and it caps the quality of what comes back. Same question, two outcomes: a vague instruction gets a vague answer, a clear one gets a precise result. This lesson teaches you to spot that difference, which sets up everything that follows.

What a Prompt Is

A prompt is the text you give a large language model (LLM) to tell it what you want¹. It can be a question, a description, a task, or some mix of the three.

A large language model is an AI system trained on huge amounts of text — "a type of language model notable for its ability to achieve general-purpose language understanding and generation."² ChatGPT, Claude, and similar tools all belong to this category. They aren't search engines. They don't look up a ready-made answer online; they generate new text that fits your instruction, working from patterns learned during training.

Here's the key point: an LLM can only understand the task through your prompt. Anything you leave out, it has to guess. The clearer you are, the closer the output lands to what you had in mind.

Vague Prompt vs. Clear Prompt

Two real examples show the gap.

Scenario 1: Asking the AI to explain a technical concept

❌ Vague prompt:

what is an API

The AI might hand you:

• a textbook definition ("API stands for application programming interface...")
• something too technical (it assumes you can code)
• or something too shallow (just "it lets programs talk to each other")

✅ Clear prompt:

I'm a product manager and I don't code. Explain what an API is usingan everyday analogy, so I can follow what my dev team means when theybring it up in planning.

The AI gives you:

• an explanation pitched at a product manager
• an everyday analogy (something like "a restaurant menu is a kind of API")
• examples tied to your actual work

What changed? The clear prompt spells out three things: who you are (a product manager), your background (you don't code), and what you'll do with the answer (talk requirements with the dev team).

Scenario 2: Asking the AI to write code

❌ Vague prompt:

write a Python function to process data

The AI is left guessing:

• what data? a list, a dict, a file, a database?
• process it how? sort, filter, transform, aggregate?
• what are the input and output formats?

✅ Clear prompt:

Write a Python function that takes a list of user ratings (1-5),filters out any rating below 3, and returns the average of what'sleft, rounded to one decimal place. Include type hints and a docstring.

The AI can produce:

• a clear function signature
• filtering and calculation logic that matches the spec
• proper docs and type annotations

What changed? The clear prompt names the input format, the processing logic, the output requirement, and the code style.

Prompt Quality Decides Output Quality

On complex tasks, how clearly you write the prompt makes a measurable difference in output quality — and since the same prompt gets reused over and over, that gap compounds with every run³. Run it the other way and a vague prompt costs you:

• More rounds of revision. You get an answer that misses, you add a detail, you get one that's still off, you add another. It takes several passes to close in on what you wanted.
• Inconsistent output. Ask the same fuzzy question twice and the style, depth, and emphasis come back different each time.
• Wasted tokens and time. A vague prompt makes the AI generate a pile of material you can't use, burning your token budget (if you're on a paid API) or your patience.

Key finding: Microsoft's Developer Tools research group found that in code-generation tasks, "prompts with explicit specifications reduced the need for back-and-forth refinements by 68%."⁴ That means you're far more likely to get usable code on the first try, instead of circling the target through a third and fourth round.

The payoff from a clear prompt shows up right away: less trial and error, a higher first-try success rate, and output you can actually steer.

Prompt Engineering Is a Skill

Prompt engineering is the systematic practice of designing and refining prompts¹. It isn't about finding one magic incantation. It's about understanding how the AI works and learning to state what you need in terms it can act on.

As of 2026, models follow instructions well — they handle complex directions, long documents, and multi-step tasks⁵. But they still lean on your prompt to set the boundaries of the task. A model can't read your mind. What you don't say, it doesn't know.

Prompt engineering covers:

• writing the basic structure of a clear instruction (Lesson 2)
• using examples to show the AI the pattern you want (Lesson 3)
• getting the AI to show its reasoning (Lesson 4)
• debugging and improving prompts systematically (Lesson 5)
• picking a strategy to match the task (Lesson 6)

This course teaches reusable methods, not one-off tricks. Once you have them, you can work more efficiently with any LLM tool.

Recap

A prompt is the instruction you give the AI, and it decides how much of your intent the AI can pick up. A vague prompt leaves the AI guessing; a clear one pins down the task boundaries, the output format, and the background it needs. Research shows a clear prompt cuts down revision rounds and gets the job done more efficiently.

The next lesson covers the basic structure of a prompt — how to break a fuzzy request into four clear parts: role, task, format, and constraints.

Next The Basic Structure of a Prompt >>