How David Perell Uses ChatGPT as a Creative Partner

Most of us are busy telling ChatGPT what to do. David Perell, however, is more focused on listening. He has his ear to the ground—and when it quivers, he responds.

Perell, you see, does not use ChatGPT to create. He creates alongside ChatGPT. Perell believes that humans aren’t good at making things from scratch. Our true genius, he thinks, lies in the art of the response. 

Perell is one of the best-known internet writers of his generation. He’s amassed almost a half million followers on X, hosts the popular How I Write podcast, and founded Write of Passage, which has taught thousands of students how to write on the internet. 

When I last interviewed Perell two years ago, we deconstructed his writing process, from random Evernote musing to polished essay. He told me that he writes from conversation. His essays were responses to his interactions with the people around him—real, warm-blooded humans who he trusted to challenge his beliefs and push him forward.

After the dawn of ChatGPT, Perell has unearthed a new gold mine of stimuli, a tireless sparring partner birthed by technology. Conversations with ChatGPT prompt his mind, often leading him down the unexpected, winding paths of the creative process.

Perell is on a mission to understand himself better and use it in his work. Nearly a month ago, on a chilly December morning, I stepped into a recording studio in Williamsburg to witness this in action. In an insightful interview, we used ChatGPT to go deep into his psyche and discover how it makes him a better writer and builder. We also delve into the meaning underlying old books, unpack business strategy, and co-write an article live with ChatGPT as our sparring partner.

These are edited excerpts of my interview with Perell. Let’s dive in.

David Perell introduces himself 

I am David Perell, and I’m a writer, podcast host, and creator of a writing school called Write of Passage. 

My number one value is the pursuit of excellence. I am drawn to being the best in the world at whatever I do. This obsession shapes who I am.

A core way I spend my time is in the company of people who are world-class in their fields, passionate people who are consistently driven to excel. A common thread I’ve noticed among them is a sense that they’re doing what they were made to do. There is a joyous fluidity, a supple graciousness to their efforts. 

Do you know what I mean? 

Think about the last time you spent time with a drummer. You’re just hanging out on a Saturday night and they're sitting on the couch, tapping their fingers, just making beats. Almost like they’re compelled to do it.

Compulsion. I think about that word a lot. I believe that all of us, carelessly walking about planet earth, are intrinsically compelled to do something. It’s in our nature, in the shape of our soul. To excel, to truly be the best, I think it’s crucial to spend time thinking about what that is. Because if you do something other than that, you’re not just fighting against your true nature, but you’re also competing with someone who is, in fact, truly built to do it. A Sisyphean endeavor if there was one. 

I believe that the path to excellence is aligning what you do with who you are. Lean in, or better still, dive in, headfirst. 

The first step is to look inward and think deeply about who you are and what you’re made to do.

Using ChatGPT to find and understand your heroes

One way to draw out the contours of your true nature is by scrutinizing what you’re aspiring toward.

We all have people in our lives that we aspire to be like. Our heroes, if you will. We are irresistibly drawn to them, but more often than not, we don’t know why. Or we’ve never been able to clearly articulate it. 

ChatGPT, it turns out, is great at distilling amorphous feelings into words. 

Over the last many months, I’ve been building a list of my heroes. It’s a list of 13 people who I look up to, along with random adjectives describing what I admire about each of them. The first name on there: “Tyler Cowen: Polymathic knowledge, consistency, fairness, curiosity, interestingness, joyous, travel.”

I input the description into ChatGPT and asked it to summarize their “vibes.” It came up with neat one-liners that tried to capture what made these individuals stand out to me. Something like this: “Tyler Cowen: A polymath with a boundless curiosity and a fair–minded approach. He embodies joy in learning and a love for travel, consistently offering interesting perspectives.”

All screenshots courtesy of How Do You Use ChatGPT?

With 13 names and their correlating details on the page, ChatGPT had a bunch of receptors to draw from. I asked it to synthesize the key overlaps between my heroes. I resonated with all the characteristics it came up with. In fact, one of them, a “balance of joyfulness and seriousness,” was something I hadn’t quite put into words before. I am a joyous person, but I’m also extremely serious about my craft. I deeply appreciate being able to articulate this. 

This brings me to something I’ve been able to articulate in the past, but have never truly explored the depths of: my core weakness. 

Learning from your core weaknesses

I can trace all the problems in my life back to one core weakness: There is a truth that I know, but I’m reluctant to express that truth because I’m worried about how it will make me look. 

This is my hamartia [fatal flaw], and I had ChatGPT conduct a motivational interview on me to unpack it. Of course, I had to remind it to ask me just one question at a time, because it has an uncanny tendency to churn out 10 otherwise. We were just two questions in when ChatGPT got pretty real. It asked me to recount a specific instance where I wanted to express the truth but felt restricted from doing so.

One of the last times I felt this way, interestingly, had to do with Dan. We had penciled in a date when we were going to record the podcast. Shortly before it, Dan abruptly sent me a big, information-dense document to parse through. I hadn’t factored that work into my week, and I had to cancel. I didn’t tell Dan why. When I did get around to working through the document, I realized that I couldn’t access the Notion that Dan had sent me. So I had to text him about it.  

I wanted to tell him that if he wanted to host a great podcast with busy guests, there couldn’t be too much prep work. If there was any, it would need to be easy for the guest. Even though I had articulated the thought and believed it to be true, I didn’t share it with my friend Dan. If it wasn’t for ChatGPT, I likely wouldn’t have.

I asked ChatGPT to summarize everything it had learned about my psychology through the motivational interview, reminding it to be as detailed and specific as possible. I thought what it deduced was very apt. 

This is not the only internal conflict ChatGPT has helped me work through.

Using ChatGPT for deep textual analysis

Even the things we read every single day hold hidden insights. All we have to do is find a way to delve deeper into them. 

I do a Bible study every day, and I’ve fallen in love with the etymology of the words used. I enjoy comparing the ways in which different translations express the same ideas. What interests me are the discrepancies. I look for them and learn from them.

ChatGPT is extremely handy when it comes to serving up translations, and I recently stumbled across a profound insight while using it for this purpose.

I asked ChatGPT for the translation of Acts 3:15 of the Book of Acts in three different ways: the English Standard Version (ESV), the New International Version (NIV), and the New American Standard Bible (NASB) versions. I’ve noticed that the ESV and NIV versions are more modern and accessible, while the NASB is more faithful to the original text.

The ESV and NIV versions translated the word archegos as “author,” while the NASB translated it to mean “prince.” An author is someone who sits in a room and writes. A prince is somebody who leads. Two very different things. I was curious to know how they could be synonyms.

So I asked ChatGPT what the meaning of the Greek word archegos was. Turns out the word archegos means author, leader, prince, or pioneer. Those are totally separate things in English. They're totally the same thing in Greek. Through ChatGPT, I could see a nuance in Greek that English doesn’t reveal.

Tracking down the etymology of this word led me to discover an unexpectedly personal insight. I am a writer wielding a pen from the confines of my desk. I am also a builder running Write of Passage in the real world. I’ve always felt a tension between these two roles—and the Greek word archegos showed me that they’re actually the same thing. What I saw as two diverging paths were fundamentally linked. Through my writing, I made a tangible impact on the real world, which I then wrote about. The author becomes the founder who becomes the pioneer who becomes the prince. This analysis, courtesy of ChatGPT, galvanized my conviction in what I do. A powerful thing indeed. 

I can take this insight forward in many different ways. For instance, I asked ChatGPT for examples of archegos in the modern world. Some of the names it came up with included Elon Musk and Steve Jobs. Finding real examples would have been much harder without ChatGPT. 

ChatGPT empowered me to understand myself. It brought me closer to figuring out what I’m compelled to do. My true nature, the shape of my soul. 

I am a writer. I am a builder. ChatGPT not only clarified my purpose; it also made me immensely more proficient in it.

Your ideal collaborators

There are two buckets of self-awareness: The first is about knowing your weakness, and the second is about finding and surrounding yourself with people who complement your personality type. 

ChatGPT already helped me delve deeper into my weakness. I discovered that it could also help me with the latter—finding people to build meaningful professional relationships with.

I am an Enneagram type three with a four-wing [Enneagram is a system that characterizes individuals into nine personality types with unique fears, motivations, and behavioral patterns, and the “wing” refers to the second, less-dominant side of an individual’s personality]. Speaking intuitively, I think I grow in the company of independent, self-confident, direct individuals. I’m inherently conflict-averse, so I don’t work well alongside people with similar tendencies. I like people who mean what they say and say what they mean. 

My guess is that I would work well with Enneagram types five [perceptive, innovative, secretive, isolated] and eight [self-confident, decisive, willful, confrontational]. Given the information it had deduced about my psychology, I was curious to know what ChatGPT would come up with. And, what do you know? When asked a pointed question about which two Enneagram types I would work best with, it came up with the same answer!

ChatGPT can show you which human beings you should work with. But it also serves as a darn good sparring partner itself. 

Using ChatGPT to set up a Socratic debate

 Sometimes the best perspective you can ask for is simply another one. Talking through an idea with someone who “gets it” can untangle the fussy knots in your thought patterns.

ChatGPT is that person: tireless, enthusiastic, and endlessly available.

As an experiment, I workshopped the New York Times’s strategy with ChatGPT. 

I started by asking ChatGPT about the biggest risks the Times faces in the next 10 years. Spoiler alert: I knew the answer would be bad. Generic questions set ChatGPT up to spew out buzzword bingo like a caffeinated college freshman in Business 101. “Climate risk,” what even!

Then, I switched gears. I asked ChatGPT a very specific question: “Look up the work of Clayton Christensen, Michael Porter, and Ben Thompson, and focus on how the New York Times can continue to shift from advertising revenue to subscription revenue.” ChatGPT did much better this time. It skipped the jargon and unearthed real insights.

A brief interlude here to say that I’m not really trying to get an answer from ChatGPT. I just try to set up some sort of Socratic debate to find the different dimensions. For example, with the Times strategy, ChatGPT has now weaved together a tapestry of knowledge comprised of the views of three business experts. I can study this and identify the fault lines of disagreement.

I even got ChatGPT to summarize the strategy recommendations we got from each expert in one sentence. Now, I have to concede that if I actually spent a whole day hashing out the Times strategy with, say, Clayton Christensen, we’d probably come up with a better answer. But that is not physically possible…this reminds me of how ChatGPT is great along three vectors: diversity (the kinds of requests you can make), accessibility (the affordability of it), and speed (the near-instant speed of responses). 

But ChatGPT isn’t just the sparring partner of my dreams. It’s also a great creative collaborator. 

Using ChatGPT as a partner in the creative process

Humans live rich, full lives colored by our unique experiences. These experiences inform our thoughts and form the essence of what we’re trying to say. As a writer, a big part of my creative process is surfacing these hidden nuggets. 

I use ChatGPT to stir my memory and discover anecdotes that form the heart of my essays. 

Let’s say I’m writing about how New Yorkers have great taste simply by virtue of the city they live in. I start by asking ChatGPT for a list of beautiful buildings in New York built between 1910 and 1933. By giving it specific instructions to search within that time period, I can use ChatGPT to pull me out of the internet’s never-ending now. I believe that true wisdom lies in things that have survived the test of time. In an age where social media algorithms direct our attention toward the latest trend, I love using ChatGPT to correct this. 

ChatGPT comes up with a solid list of buildings, the first being the Woolworth Building. 

The name of this building sparked a very specific memory. Back when I was a college student, I had a meeting with someone who worked in the Woolworth Building. We were talking about hiring, and he had a great line about how hiring is a “max function” [activities for which success is determined by the best outcome regardless of the quality of other options, unlike situations in which a balanced combination of factors is important]. What he meant was that, with hiring, you don’t need a bunch of things to come together for the action to be a success. You can interview 99 terrible candidates, but it doesn’t matter if the one that you end up hiring is talented. 

I found myself right back in that time, as a young, nervous intern. That is a powerful feeling and a vivid story. It was enough to make me decide to change what we’re writing about. We’re going to write an article about how important it is to ask yourself whether you’re undertaking a max function or a min function before getting into any work arrangement. Thank you, ChatGPT, for the beginnings of a great essay. 

Finding conversations that are food and drink for the soul

I have this idea about writing from conversations. My theory is that the mind has a tendency to get caught in grooves, and a great conversation can pull you out of those repetitive cycles. The kind of conservation that isn’t just two intersecting monologues, but rather one in which two people are in new territory, ascending onto a new plane, finding new things inside of themselves.

[Poet and philosopher] John O'Donohue says that conversations like that are food and drink for the soul. I write from these conversations, seeking them out in friends, coworkers, mentors, and strangers. Now, I can also find them in ChatGPT. 

---

Watch my whole conversation with David Perell on X or YouTube, or listen on Spotify or Apple Podcasts.

Thanks to Rhea Purohit for editorial support.