GenAI expert insights: 3 prompting techniques to know

Heya’ world! 

In previous issues, we’ve talked a lot about generative AI, covering everything from prompting 101, practical AI skills, and using it for emails to how it's changing industries like  IT, health care, and even education. Its impact is being felt in virtually every industry, though it’s not always clear how to use it productively in ways that matter to you and your workstream.

That’s why in today’s Career Chat, we’re spotlighting three GenAI prompting techniques from Dr.Jules White, whose popular GenAI courses have already taught thousands of learners on Coursera. Dr. White is a Professor of Computer Science at Vanderbilt University, has published 170+ papers, has been celebrated by the National Science Foundation, and is a National Science Foundation CAREER Award recipient.  

Recently, Dr. White sat down with us to discuss GenAI creativity, work and learning, and problem-solving. Below, we cover some of the top prompting tips he shared.    

But first, want access to an extensive catalog of AI courses (including Dr.White’s courses)? Enroll in Coursera Plus for just $25 a month from now until April 14, 2025. That’s a nearly $100 discount on more than 10,000 learning programs, expert-led courses, and job-ready certificates. 

🤔1. Use context to support problem-solving

Over the last year, as he’s spent more time teaching the ins- and outs of generative AI, Dr. White has observed the most common challenge beginners face is that they tend to treat the novel tech like an internet search. “Getting people to think about how to tell [GenAI] to solve problems rather than answer plain questions is step one. Step two is getting people away from the idea of using it to generate an answer, and instead using it to support their decision-making or problem-solving,” he says. 

Dr. White encourages users to reframe their prompts so that they’re focused more on how the output can support and improve their work rather than just do it entirely. For example, when generating an email response, Dr. White suggests you compose prompts containing important context about the recipient, such as their concerns and motivations, so that the output is balanced with your own concerns.

“The goal is really to inform your thinking…It’s something to help you arrive at a better email than you could have written before,” he notes. 

🎯2. Try the two-step “Keen Pattern”

To generate better outputs, Dr. White has found success using a technique he learned from friend and AI consultant Shawn Keen, which he adapted into a two-step approach he now calls the “Keen Pattern.” Here’s how it works: 

Step 1: Clearly define the goal and its qualities. Tell the AI what you’re doing, and then ask it to identify ways of evaluating the goal and its qualities. For example, you might ask, “What would the goal be for a really good [x]?” or “What are the ideal qualities of [x]?”

Step 2: Instruct the AI to create something based on those characteristics. After the AI has identified clear goals and qualities, simply ask it to create a version of what you’re trying to do based on what it’s just identified. 

Dr. White notes that this method not only helps clarify your goal, but also creates an opportunity to adjust and refine the output. “This helps me to start thinking more in terms of goals and outcomes I’m trying to achieve, and other things I might be trying to achieve that I hadn’t thought of,” he explains.  

👥3.  Use persona patterns

A persona pattern is a type of prompt pattern that essentially instructs the AI to take on a specific character or role, such as a detective, clown, or even a computer undergoing a cyber attack. While this might seem like just an amusing exercise (and it definitely can be), persona patterns are effective at guiding an AI’s output by narrowing the scope of its perspective to “be” something specific. 

Dr. White’s “Aha” moment came when he observed that telling an AI to “act as a computer” dramatically altered its outputs. He tested this approach in his cybersecurity class, instructing an AI to act as a computer that had been the victim of a cyber attack. It soon started to describe things that had been done to it during the cyber attack that made it respond abnormally.

“A persona can be anything. You can dream up a simulation by describing the person of the thing that you want it to simulate,” Dr. White notes. 

🤖Learn more about GenAI from Dr. Jules White

Want to learn more about GenAI from Dr. White? Explore these popular courses taught by the man himself, which are also all included with a Coursera Plus subscription (currently 25% off!):

For effective prompt engineering, try Vanderbilt University’s Prompt Engineering for ChatGPT course. Afterward, if you want to learn even more, consider completing the full Prompt Engineering Specialization.  

To enhance your life and work with ChatGPT, take Vanderbilt’s ChatGPT: Master Free AI Tools to Supercharge Productivity Specialization. Learn how to use ChatGPT for planning, project management, writing, data analytics, marketing, machine learning, and more. 

To build and train your own AI assistant, enroll in Vanderbilt’s Generative AI Assistants Specialization. Explore how to create a custom AI capable of parsing through documents, engaging in natural dialogue, and even mimicking designated communicative tones, among other things.

Okay, that’s all for today! Before you jump into your next lecture with Dr. White, let us know some of your own tips for making the most of GenAI in the comment section below. Catch ya’ later!

P.S. Want more GenAI insights? Read our full three-part series with Dr. White on GenAI and Creativity, GenAI and Learning, and GenAI and Problem Solving, or watch the full video interview. 

Have a career question you’d like us to answer next? Share it below. And, as always, we welcome YOUR advice and experience in the comments.