Where I come from, we have a saying “Answer is as good as the question”, but this wasn’t closer to the truth until LLMs appeared. Especially ChatGPT follows this rule — the answer from the model can drastically vary based on the formulation of your question, or prompt.
So let’s learn how to properly formulate a prompt, to get the most out of the ChatGPT.
ChatGPT prompting is easy, but hard to master. Let’s go over some key principles you should follow while creating a prompt that suites your needs.
We will divide this guide in couple of sections for easier understanding:
1. Key principles — follow always
- ChatGPT is a machine, so try to give your instructions with that in mind. Be precise and concise, too many information may confuse the model.
- Choose words and be precise while formulating your question, as the model needs to recognize and connect the terms from your prompt.
2. Prompting principles — follow as much as you can
In order to get the most out of the model, we have to instruct the model to give us what we need. Think of it as guiding the model by providing relevant information to get you the best possible answer.
For this to happen we want to:
- Set the model in the role
- Precisely define what we want the model to do while in that role
- Set restrictions and limitations
- Give our first input
The prompt should look like this
I want you to act as a tech reviewer.
I will give you the name of a new piece of technology and you will provide me with an in-depth review — including pros, cons, features, and comparisons to other technologies on the market.
Pros and cons should be listed in a table, features should be a bulleted list.
My first request is “I am reviewing iPhone 15 Pro Max”.
3. Cheatsheet
Below are couple of examples of what you can do with ChatGPT
