Over the last three years, artificial intelligence (AI) adoption has skyrocketed, with 50% of businesses today saying they plan to use AI across business functions in the next 12 months.
If you’re wondering how you can prepare for the integration of AI into your day-to-day work, we recommend starting with prompt engineering: an emerging skill that allows you to get the most out of generative AI. Prompt engineering involves crafting prompts to generate AI responses that are tailored to your needs.
This beginner’s guide will:
- Explain what prompt engineering is
- Review the benefits of learning prompt engineering
- Provide you with a few examples of what prompt engineering looks like
Let’s dive in. 👇
What is prompt engineering?
Prompt engineering is the practice of designing specific prompts or instructions to generate customized AI responses from generative AI tools. Prompt engineers experiment with different prompts to produce the desired response from an AI system.
A great prompt engineer will not only know how to write a prompt for a specific request—they’ll be able to create prompt templates to ensure others are consistently able to use generative AI in a productive and efficient way, helping create efficiencies across entire organizations.
After all, the more specific and context-filled your prompt is, the better your generated output will be. When it comes to prompt writing, a general rule to follow is: garbage in 👉 garbage out.
The benefits of learning prompt engineering
Whether you’re after a cushy $335,000 prompt engineering role or you simply want to learn how to better leverage AI at work, prompt engineering is well worth learning.
While learning to work with AI may seem daunting, consider that learning the foundations of prompt engineering is just as essential as skills like:
- Manipulating Excel or Google Sheets
- Communicating the details and scope of a new project or campaign
- Searching Google effectively
Learning prompt engineering can help you become more efficient at work, and can even lead to new career opportunities in the future as companies search for candidates with AI experience.
Once you get a sense of what prompt engineering looks like, you’ll also realize it might even be a little fun, since it lets you get a deeper sense of how AI works and how you can use it to further your own goals.
Examples of prompt engineering
Curious to see what prompt engineering actually looks like? Let’s dive into a few examples using Copy.ai’s generative AI platform, Chat by Copy.ai.
Designing a prompt to have an AI system summarize a news article
Let’s say you wanted Chat to summarize a news article for you, such as this BBC article on how Texas is heat-proofing its cities.
You might think that a decent prompt for Chat would be: “Summarize this article for me,” with the article’s URL included.
If you used that prompt, you’d get a summarized article, like so:
But to write a good prompt, you want to give a generative AI platform a bit more context, such as:
- What kind of audience are you summarizing the article for?
- Where will this summary be used?
- Are there any specific takeaways you’d like to highlight?
Let’s take a look at what Chat would generate when we use a more specific prompt, such as: “Summarize this article for me: [LINK]. The summary should be under 300 words and written for inclusion in a newsletter for a high-school environmental group in Austin, Texas. Focus more on the solutions included in the article.”
Now, this is what Chat returned:
As you can tell, the added context significantly shifted the copy produced by Chat. You’d see an equally big shift if you asked it to summarize the news article into a Tweet thread, a 120-word blurb for LinkedIn, or for a news anchor’s nightly news coverage script.
Crafting a prompt to generate an email in a friendly and casual tone
If you’ve ever agonized over the right way to write a tough email, you’re not alone. Fortunately, this is an area where salespeople and service professionals are already leveraging generative AI successfully: 84% of salespeople using generative AI say it helps boost sales by enhancing customer interactions, while 90% of salespeople say it helps them serve customers faster.
Here’s how generative AI can make it easier (and more efficient) to communicate with customers. For example, let’s say a customer service rep wanted to respond to a customer letting them know they weren’t able to refund their order.
They might ask Chat to write the following email: “Write an email response to a customer named John letting him know we can’t refund him on order #52104.”
Chat would return the following copy:
Not bad for a first attempt, right?
However, let’s ask Chat to write the same email in a more casual and friendly tone in order to align with brand guidelines, using the following prompt: “Write an email response to a customer named John letting him know we can’t refund him on order #52104. Use a friendly and casual tone.”
Chat would then respond with the following:
If you wanted to add in a bit more information about why the refund wasn’t available (such as the product already being used or open), you could always edit the final email produced. Often, generative AI won’t completely replace the need for you to write a single word—but it can significantly speed up the content creation process.
Need to improve your prompt but not sure how?
Simply type your prompt, click Copy.ai’s “Improve” button, and review its suggested improvements for your prompt. Edit as needed, then submit your new prompt!
Developing a prompt to generate marketing copy in a particular style
Let’s say your marketing team needs to produce application instructions for your company’s latest sunscreen product. To kick off the brainstorming process, you might want to get a few copy options from Chat.
Without thinking about a prompt for too long, you might ask Chat for the following: “Write instructions for applying sunscreen effectively.”
And Chat would deliver just that:
Those instructions, however, might be a bit too wordy and dry for your fun skincare brand.
With a bit of prompt engineering, you might come up with the following prompt: “Write instructions for effectively applying sunscreen. These instructions will be on the back of a bottle of sunscreen. The target audience is millennial women. The tone should be concise and inviting, and the instructions should be under 200 words.”
With these instructions, Chat would return the following copy:
To help ensure that Chat produces copy that’s aligned with your brand voice, you can also use Copy.ai’s Brand Voice feature. This lets you upload content examples that fit the brand voice you want generated content to align with, training the AI model to produce on-brand content for your team, all the time.
Improve your prompt engineering today
As you continue writing prompts for generative AI platforms, you should come to naturally understand the type of changes that get you the results you want. However, if you need a little help, you can always use this Prompt Improver, learn seven ways to improve your prompts.
