What the Science of Creativity Tells Us About Writing with Generative AI

This past February, I found myself in writing hell.

I had a six-week-old baby and a demanding job running marketing at A.Team. And on top of that, I was trying to write the next episode of Resignation, the TV show I co-created with my writing partner, Shane Snow.

Night after night, I’d sit in the dark on the floor of my son Max's nursery, my face glowing in the light of my Macbook.

One night, Max woke up, stared straight at me, and cried. And I knew why: I looked absolutely insane.

My face contorted in anguish as I tried to think of a half-decent joke. I looked like a Millennial version of Jack Nicholson in The Shining.

I was stuck, wasting hours, terrified that I’d let my writing partner down. And so, once I got Max back to sleep, I turned to ChatGPT to help me brainstorm. Most of its suggestions sucked, but moments later, I had the joke. Like Drano, ChatGPT had ripped through my writer’s block. I could finally move on.

From that moment, ChatGPT became my creative companion—even though I’ve never used a sentence it’s written.

When it comes to AI and writing, most people get the tech all wrong. As a writer, AI is mediocre. But as a creativity tool, it’s incredible. And the reason why lies in the science of how creativity works in the brain.

AI and the science of creativity

In 2014, researchers from the University of Pittsburgh set out to understand creativity by studying engineers developing innovative new products. They found that creativity doesn’t move in big leaps but in small steps.

“Idea A spurs a new but closely related thought, which prompts another incremental step, and the chain of little mental advances sometimes eventually ends with an innovative idea in a group setting,” they wrote.

As a writer, AI is mediocre. But as a creativity tool, it’s incredible.

For instance, a group in the study was designing a hand-held printer for kids and was trying to figure out how the door might close. One suggested it could work like a VCR tape flap. Another then suggested a garage door. Finally, a third landed on the idea of a roller door.

You see the progression below — how each team member uses each other’s analogy as a springboard to a better idea.

AI can give us unlimited ideas to jolt our creativity, and it’s already been shown to be a big boost for writers.

This summer, two British researchers ran an experiment comparing writers who worked alone to write a short story to writers who used AI to help generate story ideas. The writers who used AI as a creative springboard wrote much higher quality and more novel stories than those who did not.

Tools like ChatGPT allow us to have brainstorming sessions on-demand. It’s like having a thousand co-workers trapped inside your computer, eagerly waiting to yell ideas at you—except their ideas will probably be better. In one recent study, ChatGPT produced stronger ideas than students in a renowned innovation class at Wharton.

Of course, it’s not necessarily that important that the ideas are good. Bad ideas can just as easily catalyze those small steps of creativity that lead us to a brilliant idea.

The beauty of AI is that it allows us to take those small steps at a much more rapid pace than we could otherwise.

And even when the ideas are good, those small steps of creativity are crucial. That Wharton study also found that while AI can produce good ideas, those ideas are similar. The wildest ideas still lie in the magic hops of the human mind.