S-Tier Prompt Writing

Garbage in, garbage out.

Programmers have a saying that's older than the internet: "Garbage in, garbage out."

It means a machine can only work with what you feed it. Feed it bad input, get bad output. The machine isn't broken. The input was.

AI chatbots run on this exact rule — and now you know why. The AI predicts its answer from your prompt and nothing else. So when the prompt is thin, the prediction is thin.

The "Lazy Prompt"

Let's name the garbage. A Lazy Prompt (or lazy prompting) is a prompt that leaves out everything the AI would need to help you specifically. We'll be using this term throughout the course.

Here are some examples of lazy prompts:

Look at what's missing. Explain marketing — to whom? At what level? For what purpose? Write an email — to your boss or your landlord? Apologizing or asking for money?

You know all of that context. The AI knows none of it. And it has no way to find out — so it does the only thing a prediction machine can do with a vague setup:

It predicts the average.

The most generic, middle-of-the-road, written-for-everyone response that fits your words. That's why lazy prompts all get back the same flavor of junk: technically correct, completely useless to you. It's not an accident. It's math.

Vague in, vague out. Specific in, specific out. The quality of the output is decided before you hit enter.

One important note: Can a bad, lazy prompt sometimes give you a good response? Yes, sometimes you get lucky. But good prompts guarantee good responses. We're here to remove luck from the equation.

Which means the fix was never "find a better AI." You're holding the same tool the power users hold. The fix is upgrading what goes in.

But before we start upgrading, there's one more distinction you need — because "a better answer" isn't actually what you want most of the time. There's something better than answers.