Get the Most Out of AI with the Right Prompts
Artificial intelligence is not an unknown at One North. Not only have we published more than half a dozen blog posts on the topic, but when it comes to the newest shiny AI object, Generative AI (like ChatGPT), some of us have even used it in our work.
Being the conscientious digital experts that we are, we tread cautiously into these AI waters, testing and learning as we wade deeper into the fast-moving currents of Gen AI. We’ve long understood that we need to use AI responsibly. Now, as AI grows in acceptance, and both our colleagues and clients are increasingly interested in using this new tool, our focus must shift to using AI effectively.
Before we move on to exactly how we can do that, let’s set some ground rules. When I reference Gen AI, I’m talking about the large language models (LLMs) like OpenAI’s ChatGPT and Google’s Bard. While there are plenty of AI tools for other applications, like Midjourney for imagery, I’m going to stick with the text-based chatbots for this post.
Garbage In, Garbage Out
While it’s a cliché at this point, the old computer science adage of “garbage in, garbage out” remains the key to getting the most from Gen AI. The quality of the output is only as good as the quality of the input. That means the prompts we give our AI are extremely important—and the better we are at creating high-quality prompts, the more likely our AI will deliver high-quality results.
Prompt writing is easy, but it can be a bit tricky to master. I like the way Anna Bernstein, a prompt engineer at Copy.ai, explains it. She likens prompt writing to a wizard casting a spell: “If you say the spell slightly wrong, something slightly wrong can happen.”
Prompt Like a Pro
Given the importance of prompting, I combined what I’ve learned from my own experience using AI with advice from the pros to develop a handy list of tips for writing effective prompts:
- Treat the AI Like a Person. Be conversational and unafraid to ask follow-up questions. Ask a question, get a response. Based on that response, ask another question, etc. You wouldn’t just ask your friend one question and walk away; you’d have a conversation. Do the same with the AI you’re using.
- Use a Thesaurus. Sometimes you need to find the right words to elicit the right response from the AI. If your first prompt doesn’t result in what you’re looking for, ask again with slightly different wording.
- Provide Context. Providing relevant background information can help the AI understand your intent, and intent is a key area of focus for these chatbots. The better it can understand your intent, the more accurately and completely it can answer your query.
- Challenge the Responses. After you’ve received an answer to a question, challenge the AI to justify its answer. You can ask follow-up questions like, “Why do you think that?” or “What evidence supports your answer?” This will often result in some specific sourcing information that you can use to verify the accuracy of the response.
- Choose Your Words Wisely. Just like any good marketer, AI chatbots appreciate direct, concise writing. Try to use active verbs with a little more punch. Telling a chatbot to “summarize this,” is more powerful than, “provide a brief overview.”
- Create an Identity. A Gen AI bot can write from various points of view and can assume the identity you give it. Tell it to write the answer as if it’s a professor or CEO. You could also have it answer as if it’s Ernest Hemingway or Stephen Hawking, or even respond in the style of Quentin Tarantino. (Remember to exercise judgment about what is right for your particular application. These are just examples.)
Of course, this is just a handful of tips, and there are many more things to consider depending on your particular needs and situation. Along those lines, I did find a helpful little “Cheat Sheet” from marketing and AI consultant Zain Kahn for prompting ChatGPT that covers specific instances for marketers, sales, designers, coding, and research:
As businesses continue to expand their understanding of how to use tools like Generative AI and slowly fold them into work processes, it’s essential that they develop best practices for using them responsibly, effectively, and efficiently. By understanding how to better prompt the language models you are using, you will get higher-quality, more accurate results in a shorter timeframe—leading to greater workstream efficiencies.
Photo Credit: Saj Shafique | Unsplash
Brendan O’Neill is a Content Strategy Lead at One North. He helps clients evaluate and reimagine content structure, development, tactics, and strategy incorporating both industry best practices and innovative methodology. Brendan has an extensive background in journalism, editing, and managing content for everything from local newspapers and trade publications to national consumer magazines and Fortune 500 brands.
Most unusual job: Wrote a book on pro wrestling trivia. Well, “compiled” is a more accurate description, as I’m far from a pro wrestling expert.
What a superpower you wish you had: Time travel. It’d be amazing to go back and forth through time…and I’d be tempted to change a few things in the past, which might be dangerous.