One of the big promises of AI is that it will democratize writing—and the way the frontier models continue to develop seems to support that.
ChatGPT’s Canvas allows you to collaboratively edit documents with (and within) the AI, in much the same way you would with an actual human editor.
It gives you helpful suggestions for how to make your writing clearer, proposes places to insert examples that drive home your points, wordsmiths the copy, and gives you recommendations that are, more or less, pretty good.
The beauty in all this is that for people to whom writing is a chore, LLMs are tools that unlock new ways of interacting with the world.
They make something that we writers take for granted, by which I mean the ability to clearly and effectively communicate an idea, available to everyone, at any time, for no cost at all.
I can see the incredible value and utility in that—and the potential it has to help people.
Imagine for a moment what it would be like to look for a job if English wasn’t your first language. Every form of written communication would be filled with risk.
Risk that you’d misspelled a word.
Risk that you’d used language incorrectly.
Risk that you’d seem unintelligent.
Now, with an LLM’s help, your cover letter, resume, and emails to recruiters become tools that highlight your strengths rather than barriers holding you back.
Using an LLM significantly boosts your chances of reaching the interview stage, opening doors that previously might have been closed.
Which is definitely a good thing.
And yet, I still have my concerns.
While I certainly acknowledge the tremendous value LLMs bring, I think it’s also important to consider what we might lose when we rely too heavily on them.
I’ve sounded this alarm before, so I don’t want to cover the same ground again, however, I think it’s worth unpacking what we mean when we say “writing.”
Because not all writing is the same.
A LinkedIn post, a business insight article, a personal essay, a novel, a poem, and an email are all forms of writing, but they’re different in almost every way.
Yes, they’re all taking an idea and communicating it to an audience, but the cognitive work and creativity required to produce each one are different. And so are the goals, which means the way you approach each literary form should be different too.
I probably don’t need to tell you which ones are okay to use an LLM for and which ones are not because all you need to do is ask yourself this: If I knew an LLM helped produce it, which ones would I be okay reading?
To be clear, as a writer, I have no issue using an LLM to help me refine a piece of thought leadership like this one. I have no issue using an LLM to refine an article. And I certainly don’t have an issue using an LLM to help me refine a point I’m trying to make.
However, the key word here is refine.
As I’m fond of saying, before an LLM can write for you, you have to write for it.
Marketer and analytics veteran Andy Crestodina says that until you add your context and your personal perspective, AI stands for “average intelligence” and “assume incorrect.”
He’s right.
Why? Because when LLMs were trained, they essentially “ate the internet,” which means that without context and perspective, without your personal viewpoint, LLMs simply assemble the average of what it found on the World Wide Web.
To be sure, it’ll sound good, and you’ll be impressed with it, but it won’t be unique, and it certainly won’t be yours.
This matters because your personal experiences and insights infuse your writing with depth and authenticity that no AI alone can replicate.
This doesn’t matter with all writing, but it does matter with some writing.
And it’s definitely worth considering.
What makes your writing different from mine is the experience and knowledge you bring to it.
Consultant and former Hubspot Head of Content Jay Acunzo puts it this way: “Both A.I. and humans are powered by LLMs. A.I. is informed by large language models. You are informed by little life moments.”
That’s an incredible insight and you’d be wise to take it to heart.
Yes, LLMs can dramatically lessen the time it takes to go from “write” to “have written,” but if that’s your goal, if that’s all you’re trying to do, then you’re focused on the wrong thing.
In many ways, “to write” is the whole point, because to write is to think.
Writing forces us to slow down and articulate our thoughts. It makes us draw on our experiences and grapple with complexity. It makes us think back on our experience and mine it for, as Acunzo says, those little life moments.
Perhaps the great promise of AI isn’t actually to democratize writing, but to give everyone a way to more deeply explore their experiences.
Because that’s the kind of writing I want to read—and, hopefully, the kind we all aspire to create.