Unlocking the Potential of Generative AI: 10 Prompt Engineering Recommendations
In the ever-evolving landscape of technology, Generative AI has emerged as a groundbreaking innovation, reshaping the way we approach problem-solving, creativity, and design. As product managers and designers, it's crucial to understand the concept of prompt engineering – the art of effectively communicating with AI models to elicit desired outcomes. This article delves into the essence of prompt engineering and provides best practices for harnessing the potential of Generative AI.
Prompt engineering is the process of crafting inputs (prompts) for AI models in a manner that maximizes the likelihood of achieving the desired output. It's akin to speaking a language that AI understands fluently, ensuring that your requests are interpreted accurately and effectively.
Prompt Engineering Best Practices: Prompt engineering expertise comes with practice, but here are starting recommendations to leverage to make your prompt engineering more effective:
Use clear language. Your prompt should be easy for the generative AI model to understand. Avoid using jargon or technical terms that the model may not be familiar with.
Be verbose, but break things into steps. Oftentimes, providing a more verbose prompt, as long as it is clear and understandable, will lead to a better result. For example, indicating “outline a blog post written in a professional manner to an audience of marketing professionals. The blog post is geared towards technology leaders and is driven around risk management and how to monitor potential impacts. This article should provide recommendations and risks to be aware about.” is a verbose prompt that should get better results than a more basic prompt “write a blog post to technology leaders on risk management.” When being verbose though, it also helps breaking your prompts into steps. This helps ensure there is clarity on the steps desired.
Define your AI assistant. Give the AI assistant as part of the prompt context on how it should respond. For example, maybe you are looking to have AI edit an article that you wrote. Provide as part of the prompt something like “You are an expert editor who has decades of experience providing critiques to product managers and designers on their writings. You do not suggest things that you are uncertain about.” this might provide improved output. For some models, you can update the ‘system prompt’, which provides the context of the model as a precursor and where you would typically define your assistant.
Define your audience. When it is relevant that you are doing research or writing to a specific audience (e.g., product managers), then provide the AI with the context that your audience is product managers.
Define your goals and objectives. Providing a generative AI model your goal or objective in a clear manner. For example, if asking a generative AI model to write a first draft, you may provide a goal of the style of writing (e.g., professional), you may provide a desired format, and you may provide a desired word count.
Be specific. The more specific you are with your prompt, the more likely you are to get the results you want. For example, instead of saying, "Write a story about a character who goes on an adventure," you could say, "Write a story about a character who goes on an adventure to find a magical treasure."
Provide examples. One way to enhance your AI models is by providing an example or two of what to expect. This will help provide the AI model context on the problem and format asked.
Give the model some creative freedom. Don't be afraid to let the model go off on its own a little bit. Sometimes, the best results come from unexpected places. Many models have a temperature setting where the lowest temperature is a deterministic model where the most probable output is repeatedly provided back, and the highest temperature setting is a highly stochastic model where you get a variety of results.
Understand the model's limitations. Familiarize yourself with the strengths and limitations of the AI model you are using. This knowledge will help in setting realistic expectations and crafting effective prompts. Models change rapidly, so don’t get stuck in one model. Be flexible by trying different AI models for different tasks.
Test your prompts. Once you've written a prompt, test it out with the generative AI model to see what results you get. This will help you fine-tune your prompts and get the best possible results.
If you are looking to learn more about prompt engineering, then there are many resources out there. One course that I have taken is Jules White, Ph.D. of Vanderbilt’s course titled “Prompt Engineering for ChatGPT,” which is available on Coursera. Of course, you could also try out asking different AI models about best practices around prompt engineering and see what they provide. In fact, when starting this blog post, that is exactly what I did with a few different AI models, and then I combined and edited what is posted here today.
Prompt engineering is more than a technical skill; it's a creative and strategic tool that can unlock the vast capabilities of Generative AI. By mastering this art, product managers and designers can revolutionize their approach to problem-solving, innovation, and design. Remember, effective communication with AI not only enhances efficiency but also opens doors to unparalleled creative possibilities. With a little practice, you'll be able to create effective prompts that will help you get the most out of your generative AI model.
Happy prompting!