What is Prompt Engineering?
In layman terms any text input you are putting in ChatGPT/Claude/Gemini etc. is a prompt.
Hence, Prompt Engineering is the process to optimize this prompt to ensure that we use minimum tokens to get the best possible output.
Prompt Engineering has been deeply studied since LLMs became mainstream and there are detailed guides on it like this highly detailed and meticulous www.promptingguide.ai. There are many more such guides and survey papers which you can read about.
But in most practical use cases you will find these 5 principles of prompting common among all.
These principles have also been highlighted in the book - Prompt Engineering for Generative AI ~ James Phoenix & Mike Taylor
Overview of the 5 Principles of Prompting:
Following are the 5 Principles of Prompting, we will deep dive into each of them -
- Give Direction - Describe the desired style in detail or reference a relevant person
- Specify Format - Define what rules to follow and the required structure of the response
- Provide Examples - Insert a diverse set of test cases where the task was done correctly
- Evaluate Quality - Identify errors and rate responses, testing what drives performance
- Divide Labor - Split the task into multiple steps, chained together for complex goals
We will now go deeper into each of these principles and see a bunch of examples to clearly understand what they mean.