AI prompt engineering
A blog post by Daniel Miessler with a few techniques for increasing the quality of responses from AIs.
Persona
- Tell the AI who you want it to be
- "you are an X with 'defining characteristic y"
Format
- Tell the AI what output format you want
- "you produce bullet points with no more than 12 words", "you produce valid JSON", etc
Task
- Give it the thing you want done
- "Write a summary of this article", "Create a blog post about Joe Blogs using their linkedIn and social media accounts"
Steps
- Give the steps you want it to take to complete the task
- "Break down this input from a journalism standpoint, then look at its entertainment value, and then evaluate its humor level"
Output
- Tell it exactly how you want the output to look
- "I want three sections: Introduction, followed by 2 paragraphs of analysis, Main Points, followed by 1 set of 5 bullets, and Takeaways, which gives you 3 things to immediately start doing tomorrow"
Examples
- Show 1 - 5 examples of ideal output
- "Here’s an example poem that you should try to capture the feel of in your responses", "The JSON should have this exact format", "The story should have this character to it"
Tweaking
- Tell it what to include or what to exclude
- "Don’t start with a preamble sentence; just create the output", "Make sure you always capitalize the output"