The explosive growth of ChatGPT to 100m users in a few short months has brought AI to the mainstream. That means it’s becoming a marketable skill to be able to work with AI effectively, and one that people are already hiring for.
The growing discipline of “prompt engineering” is the process of discovering prompts – i.e. the text we input to ChatGPT – which reliably yield useful or desired results.
The great American writer Mark Twain wrote “The difference between the almost right word and the right word is really a large matter—it’s the difference between the lightning bug and the lightning.”
Nowhere is this more true than in prompt engineering, where knowing the right way to ask can yield immeasurably better results.
“The difference between the almost right word and the right word is really a large matter—it’s the difference between the lightning bug and the lightning.”
– Mark Twain
As the author selected by O’Reilly Media to teach Prompt Engineering, the creator of the Vexpower Prompt Engineering courses, I wanted to share 7 places I look to for inspiration.
Table of Contents
1: Meme Mapping
In order to find the right combination of words and phrases to use with AI, all it often takes is just seeing more examples in a space than anywhere else until you start to notice patterns.
For example how James Clear noticed that many best sellers used the same pattern, pairing an unexpected descriptor word with the topic of their book: “Extreme Ownership”, “Deep Work”, and Clear’s own entry, “Atomic Habits”.
Meme mapping is a structured process for doing this systematically and at scale. First you choose a group or channel to target, and build a swipefile of successful examples.
Then you use inductive coding – labeling the data based on what you find in looking at it – to identify recurring patterns or ‘memes’ that seem to work.
Say for example you needed to write a blog post in a competitive industry. Use meme mapping to identify what search terms and article features are common amongst posts that rank on the first page of results on Google, so you can ask ChatGPT to incorporate those elements in your own work.
2: PromptBase
It says a lot about the speed of adoption of AI that there’s already a marketplace for buying and selling prompts. Primarily starting out focused on AI art, PromptBase has since expanded to ChatGPT and GPT-3.
Once you pay for your prompt, you gain access to it and can copy and paste it across to the ChatGPT interface and hit submit. Usually these prompts are ‘templates’ in which you need to replace key words or phrases with your own, to adapt it to your use case.
3: OpenAI Cookbook
ChatGPT might have just gone mainstream, but software engineers have been working on AI alignment for years. Even if you don’t know how to code, you can click into various repositories in GitHub – like the OpenAI cookbook – and get an idea of how these things work.
Thankfully, prompts are always in plain text, so you can ignore the code and just see what they recommend when trying to accomplish different tasks and goals.
Usually these are the .ipynb files (Python Notebooks) which you can load directly in GitHub, or in Google Drive using Colab, without setting up a coding environment.
4: AI Twitter
This is a case where wasting time on social media can actually be productive! Many of the leading proponents of the AI revolution are regular Twitter users, and it’s often where news is broken first before filtering down through mainstream news organizations and groups.
Prompt engineers like Riley Goodside – the first person with “prompt engineer” in his job title – share generous insights from their own work.
Rather than provide you with a list of users to follow – the field moves far too quickly for that – I recommend starting by doing a few searches, following those who stand out as interesting, and liking a few of their posts.
Before long, the algorithm will start showing you AI content, and in no time you’ll start finding talented prompt engineers that I haven’t even found yet.
Conclusion
Prompt engineering is an in-demand skill in a rapidly changing industry. In order to stay on top of it you need to make finding inspiration for new prompts a priority.
From meme mapping to identify what words and phrases to prompt the AI with, to buying prompts on PromptBase, digging through coding notebooks on GitHub or browsing AI Twitter, there’s plenty out there to make sure you never run out of good ideas for prompts again.
Related posts:
- Consulting for Change: How Change Management Experts Drive Growth and Innovation - September 8, 2023
- ChatGPT Prompts – 4 Sources of Inspiration - April 7, 2023
- The Savvy Marketing Agency Guide to Wild Growth in 2023 - January 23, 2023
