When ChatGPT launched in November of 2022, I, like most people in the business community, was unsure what to make of it.
There was no instruction manual, no established best practices, no obvious use case beyond asking it to write something and then sitting back and watching while it did exactly that.
No matter how obscure your subject was, no matter how little information you gave it, the machine whirred to life with alarming speed and produced a piece of writing that was often better than what you could have written yourself.
As a writer, this immediately piqued my curiosity and had me asking a perplexing question: In a world where writing is fully automated, and when we no longer need to write at all unless we choose to, will we?
It’s a question worth grappling with because writing, while difficult and time-consuming and frustrating, is also difficult and time-consuming and frustrating in exactly the way it needs to be.
This is because writing is how we refine our ideas. It’s how we understand and challenge our perspectives. And by sharing these ideas and perspectives, writing is how we shape the world around us.
In a very real way, to write is to think, and to think is to be human—and what it means to be human, however you define that, is a topic we’re grappling with now more than ever.
In an essay from March 2023 titled “The Upside—and Downside—of Arguing with ChatGPT,” I wrote that ChatGPT was a tool that could allow us to spend more of our time on more of the things that matter, like revising our writing—by which I meant: revising our thinking.
“We call it revision,” I wrote, “because it allows you to ‘re-vision,’ to reimagine what you wrote so you can get closer and closer to what it is you’re really trying to say.”
And what I’m really trying to say right now is that writing is important and fundamental to who we are. Writing is thought made manifest, and it’s the reason why, in graduate school, I studied the essay—that literary form whose very name, from the French essai, means to “try out” or “to attempt.”
What I’m attempting to understand here, what I’m essaying, is what we gain and what we lose if we abdicate our writing, by which I mean our thinking, to a machine. (I’ve wrestled with this idea before, here and here, and I imagine I’ll return to it in the future.)
I’m certainly not the only one thinking about this, though. In fact, in a recent essay called “Writes and Writes Not,” famed computer scientist and Y-Combinator co-founder Paul Graham also considers the future of writing.
“The reason so many people have trouble writing is that it’s fundamentally difficult. To write well you have to think clearly, and thinking clearly is hard. And yet writing pervades many jobs, and the more prestigious the job, the more writing it tends to require.”
He goes on to declare that the world will soon be divided into two classes of people: those who write and those who don’t.
The division, though, isn’t really about writing. Rather, as you’ve probably guessed by now, it’s about thinking—and about what happens when one group thinks and one group doesn’t.
Graham points out that while we’ve survived the loss of many traditional crafts (there aren’t many blacksmiths left, he notes), writing is fundamentally different. A blacksmith shapes metal; a writer shapes thoughts.So if we lose our ability to write, we lose a unique form of cognition essential to who we are.
Why struggle through a concept when we can literally push a button and fast forward to the end, to the point where the thought is fully formed?
The great irony here is that I’m writing this essay in dialogue with Claude, Anthropic’s frontier large language model, in the same way I would dialogue with an editor or writing group.
In fact, it was Claude who suggested I surface this irony—a suggestion I found quite good—but it brings with it a paradoxical thread I’ll need to resist pulling, at least for now.
Because what I want to focus on is this: it’s important that we’re mindful of what we’re giving up when we use LLMs.
They’re useful and have fundamentally changed the business landscape, but we need to know when to resist the siren’s call of this new technology, when to avoid the easy path, so we can engage in the necessarily difficult work of writing and thinking.
Before an LLM can write for you, you have to write for it. I believe that wholeheartedly. LLMs can be incredible developmental editors and copyeditors and proofreaders and sounding boards, but—and this is the important part—they should be revising your work, and not the other way around.
Is this the end of writing? Or the end of thinking? It’s not for me, and it doesn’t have to be for you.
There’s a choice to be made here, and it’s worth some real consideration. One allows us to take the easy path, to skip the hard stuff.
The other asks us to embrace the struggle of being a human being, in all its frustrating complexity.
I’m excited to see which one you make.