The Secret to Using GenAI Tools Effectively: Insight from Coursera’s CEO

Jeff Maggioncalda has been a Silicon Valley CEO for over 25 years, and the CEO of Coursera since 2017. He holds a Master of Business Administration (MBA) from Stanford Graduate School of Business, which he obtained after earning his bachelor’s degree in Economics and English from Stanford University. As far as resumes in the tech sector go, Maggioncalda’s is certainly impressive—but there’s one pivotal role, long removed from his resume, that started it all: graphic artist.

After an unsuccessful stint as a restaurant server, Maggioncalda, then an undergraduate student, needed a job, so he turned to his local newspaper, the Stanford Daily. That’s where he found the lucrative opening, advertised at $12.50 per hour. He wanted that salary; the only thing standing in his way was his lack of design experience.

But do you know what he did have? A computer, a bootlegged copy of Adobe Illustrator 88, and access to his university’s then state-of-the-art LaserWriter printer.

Maggioncalda quickly taught himself this early graphic design software, put together a portfolio, presented it to the hiring team, and got the job. “Somehow that was always my personality: I would always learn the newest technology just enough to add value to somebody—never enough to actually be the best at it—and then, once everybody else learned it, I'd learn the next new thing.”

Fast-forward to the end of 2022, and you can guess what Maggioncalda’s latest technology obsession became: Generative AI and, more specifically, ChatGPT.

“The first time I got a response, I was blown away,” he recalls. “I just used it nonstop all through December—that’s all I did; I just banged on it and banged on it. I said, ‘What is this technology? It’s ridiculous.’”

Since then, he’s been a fast learner and vocal supporter of the ways ChatGPT and other large language models (LLMs), like Google Gemini and Claude by Anthropic, stand to change the way we work. As Coursera’s CEO, he’s leading the company toward an AI-powered future, determined to use the technology to further the company mission: To empower anyone, anywhere to transform their life through learning.

But here’s Maggioncalda’s secret: The skill that will enable you to learn this technology enough to add value—but, crucially, not enough to be the best—has nothing to do with AI. It’s called systems thinking.

Remember, Maggioncalda’s secret only works if you want to use AI tools to enhance the way you work. If you want to work in AI, for example as an AI developer, AI engineer, or AI researcher, you will need to master the technical aspects of AI. Read more about technical AI careers.

What is systems thinking?

Systems thinking is the ability to break down processes into their components and outcomes, identify relationships between the components, and use those relationships to influence the outcomes. Here’s how Maggioncalda describes it:

“Systems thinking is the ability to see a number of things that seemingly are unrelated, and then be able to determine which things are related to each other and which ones are not. When they're related to each other, how are they related in a functional way such that, if I'm trying to produce an output and I have all these different pieces, if I arrange these pieces in this way, the system produces this output more effectively and efficiently.”

Systems thinking example

Throughout his time as a CEO, Maggioncalda has identified systems thinking as a key skill among top business leaders. At the highest level, leaders think about their business as a system, and they make decisions based on that system. For a more tangible example, let’s narrow the scope and focus on a job within that system. Here’s Maggioncalda's framework for applying systems thinking to jobs.

First, he starts with the question, “What is a job?” According to Maggioncalda, a job is:

• The results that you’re responsible for producing

• The deliverables that produce those results

• The tasks, activities, and actions you take to produce the deliverables

• The decisions you make to perform those tasks, activities, and actions

• The people you depend on to perform the tasks, activities, and actions

• The people depending on you to perform the tasks, activities, and actions

• The tools you use to perform the tasks, activities, and actions

• The managerial support you need to perform the tasks, activities, and actions

It’s important to notice here that it’s not just the pieces that make up the system, but the way those pieces are connected to one another. 

Read more about applying systems thinking to your job role in the article How to Use GenAI in Your Job: Insight from Coursera’s CEO.

Now, what does systems thinking have to do with LLMs?

Don't start with the technology. Start with the problems that you can solve, then understand enough about the technology to understand what's feasible. — Jeff Maggioncalda, Coursera CEO

Systems thinking and ChatGPT

To understand the link between systems thinking and ChatGPT, it’s important to be familiar with the types of tasks that these GenAI tools are best suited for: reading, writing, and learning tasks. For example, you can input a research study and ask the LLM to summarize the key findings.

However, these tools are not optimal search engines, meaning they’re not designed to answer any random question with no context. They’re trained on some data, like text from the internet, which helps them formulate answers to random questions, but their answers aren’t always reliable. For more reliable answers, as a user, you must provide the LLM the context it needs to effectively and accurately answer your question.

In essence, LLMs need you to provide them with information about the system in which your question exists. How, exactly, do you do that? Let’s turn back to Maggioncalda to find out.

Is AI dangerous? Is ChatGPT safe? These questions are on many people’s minds as this technology becomes more prominent. It is important to be aware of AI safety and security, and Maggioncalda acknowledges that he has some fears about the unintended outcomes of quick integration or the way bad actors may eventually use AI. For Maggioncalda, it helps to conceptualize separate AI use cases. “There's a little bit of a difference for me in my personal relationship to the technology, like Jeff and AI, and then there's Coursera and AI, and then there's the world and AI,” Maggioncalda says.

How to use ChatGPT with systems thinking

If you want ChatGPT, Google Gemini, or any LLM to perform a thinking task—in other words, to help you answer a question, figure something out, or solve a problem—you can derive more accurate responses if you provide your systems-based approach as context before asking your question. “The ability for it to help me think through things, to formulate rational patterns of thought was stunning to me,” Maggioncalda says.

Here are his three steps for using LLMs as a thought partner:

1. Detail your system.

Before you open your GenAI tool, consider the system around the problem you’d like to work through. To try this yourself, start with Maggioncalda’s framework above for thinking about a job role and customize it for your role or team. This will be your context.

2. Provide your context.

Now you’re ready to consult your LLM. Input your context into your GenAI tool using Maggioncalda’s go-to prompt:

Consider the following CONTEXT and reply "I understand the context" but do not explain:  

<<< BEGIN CONTEXT >>>

[add context here]

<<< END CONTEXT >>>

In this prompt, you can change the word CONTEXT to describe your context. For example, with a job role, you may write, “Consider the following job role.”

“Once you tell the LLM, ‘This is what a job role is and here's a particular job role,’ you can then ask it, ‘How are the results going to change [with the introduction of GenAI]? How might the decisions change? How might the activities change?’ So you have a framework, a set of principles that you could use this against,” Maggioncalda says, bringing us to the final step.

3. Instruct the LLM.

Now, you’re ready to ask your question. Depending on the context you provided, here are some sample prompts to try:

• For ideation: Use the CONTEXT to generate [output] about [topic] with the goal of [goal].

• For summarization: Summarize the key points of CONTEXT focusing on key aspects and implications for [goal].

• For systems analysis: Analyze the CONTEXT by examining its components and their relationships.

• For enhancement: Advise me how to improve CONTEXT in order to better [goal].

Prompt like a pro

Ready to go deeper on prompting techniques with Coursera CEO Jeff Maggioncalda? Try our guide, Useful Generative AI Prompt Techniques for Everyday Work, to grow your AI skills, better integrate GenAI into your processes, and prepare for your future job role.  

Keep learning with Jeff Maggioncalda on Coursera

In his course Use Generative AI as Your Thought Partner, Maggioncalda breaks down the process for using ChatGPT with systems thinking, with your career as an example. In the second article in this series, How to Use GenAI in Your Job: Insight from Coursera’s CEO, Maggioncalda further breaks down this example, narrowing the scope to your role. In the third article in this series, How to Use GenAI to Advance Your Career: Insight from Coursera’s CEO, Maggioncalda shares his tips for using a similar process to advance your career.