Everything you type into ChatGPT is a prompt. It's the AI's only clue to what you actually want.
You’re going away for the weekend and a friend is minding your cat. If your note just says “Look after the cat”, they’ll have to guess. Maybe they feed her twice a day, maybe not at all. If it says “Half a pouch at 8 am and 6 pm, fresh water every day, she hides under the bed when the doorbell rings”, they’ll get it right.
An AI works the same way. A prompt is the note you leave it: whatever you type to tell it what you want. It can’t read your mind, so the note is all it has.
What makes a good one
- The job: what you want done, like “summarise”, “write” or “explain”
- The background: anything it needs to know first
- The reader: who it's for, like your boss or a ten-year-old
- The shape: how long, what tone, bullet points or a paragraph
It’s a conversation
You don’t have to get it right the first time. If the answer is off, reply with what to change: “shorter”, “less formal”, “add an example”. Every message you send is a new prompt.
What it isn’t
It isn’t a magic spell. There are no secret words that unlock a hidden mode. A clear, specific request in plain language works better than clever tricks.
Where the comparison breaks down
Your friend can text you if the note is unclear. An AI usually won’t ask. It fills the gaps with its best guess and carries on, which is why the detail matters.
How the words fit together
You write a prompt. The LLM breaks it into tokens and writes a reply piece by piece. If it doesn’t know the answer, it may make one up: that’s a hallucination.
Say what you want, who it's for and how long it should be. “Write a short, friendly email to my landlord about the broken heater” beats “write an email” every time.

When you'll hear it
“Try changing the prompt.” Ask it in a different way and you'll probably get a better answer.
“That's a good prompt.” That request was clear enough for the AI to do a good job.
“We need some prompt engineering.” We need to get better at wording our requests so the AI gives better results.
“Can we share the prompts that work well, so everyone gets the same quality?”


