Prompt Engineering Course by OpenAI — Prompting Guidelines
Get to know OpenAI’s Guidelines for Better Prompting
--
OpenAI and the learning platform DeepLearning.AI just released a course on Prompt Engineering. The course is supposed to be free for a limited time, which is why I decided to do a series of articles on the take-home messages from the course, including only the most important examples and some additional information to help you boost your interactions with ChatGPT.
The course is mostly oriented to developers or ChatGPT API users, and it is focused on helping you generate better ChatGPT outputs for your custom application. Nevertheless, quite some of the prompting techniques can be also used when interacting with the chatbot by using the web interface.
In this first article, we will review some of the OpenAI guidelines for prompting that allow you to maximize ChatGPT’s output. These guidelines can be summarized in two principles: be clear and specific and give the model time to “think”.
The theory is clear, but let’s see how to apply those principles in practice!
Principle I: Be Clear and Specific
This first principle is about giving clear and specific instructions to the chatbot. This best practice will reduce the chances of getting irrelevant responses. By being clear and specific one can guide the model towards the desired output.
A specific prompt doesn’t mean a short prompt. Details can make the prompt more clear and more specific about the desired outcome. Longer prompts provide more clarity and context so that the chatbot understands the task to carry out.
OpenAI recommends using the following four tactics to be as clear and specific as possible.
#1. Use Delimiters
Writing clear and specific instructions is as easy as using delimiters to clearly indicate distinct parts of the input.
This tactic is especially useful if the prompt includes pieces of text. For example, if you input a…
