If you haven’t already, sign up for The AI Exchange Newsletter where we’ll continue posting helpful resources like this!

This is a practical guide on how to generate better AI outputs by up-leveling your inputs

Created by: The AI Exchange Team

What is “prompt engineering”?

A “prompt” is the input that guides a generative AI model to generate useful outputs. Generative AI tools like ChatGPT, GPT-3, DALL·E 2, Stable Diffusion, Midjourney, etc. all require prompting as their input.

Source: https://docs.cohere.ai/docs/prompt-engineering

In a natural language processing (NLP) context, “prompt engineering” is the process of discovering inputs that yield desirable or useful results. As is the story with any processes, better inputs yield better outputs; or commonly said another way “garbage in, garbage out.”

Source: https://www.youtube.com/watch?v=1NQWJjgi-_k

Designing effective and efficient prompts will increase the likelihood of receiving a response that is both favorable and contextual. With a good prompt, you can spend less time editing content and more time generating it.

Going from beginner → advanced prompt engineer

As companies like PromptBase arise around the idea that the prompt is the “secret sauce” to using generative AI, prompt engineering could easily become the “career of the future.” But, any generative AI user can become an “advanced” prompt engineer. Here’s how:

Spend time with the tools

• The more time you spend asking ChatGPT questions and receiving responses, the better your idea will be of various prompting approaches and their individual strengths and weaknesses

• Use Open AI’s GPT playground to perform interactive trial and error with variations in your prompt, model, temperature and top_p values (uniqueness of answer, i.e. creativity), and more available within the UI itself.

  

  Source: https://beta.openai.com/playground

Use your expertise in other areas to guide your prompting

• If you’re already a subject matter expert in something, consider figuring out how to apply your personal skills to generating the best prompts in your field
• For example, if you’re an expert in SEO, what questions do you ask yourself when creating SEO strategies? How can you translate this knowledge into better prompts to generate the same level of output with AI?

Become a prompt researcher instead of engineer

• The term prompt engineer glosses over the idea that prompt formulation takes hypothesizing, research, result measurement, and repetition. Instead, approach prompting like a research project.
• Try as many different variations and formulations of your prompt as possible. One problem can have hundreds of solutions and one solution can have hundreds of approaches. The same can be said of prompting.