Remember this?
“Blogging: Never before have so many people with so little to say said so much to so few.”
Thank you, Despair Inc.
AI content generation has told blogging to hold its beer.
At the moment, it seems nearly impossible to open LinkedIn without finding someone talking about AI writing. AI slop. How to detect it. How not to sound like ChatGPT. Whether AI-generated content should be labeled or watermarked. Whether using it is dishonest. Whether people can tell. Whether em dashes are now evidence of moral failure.
I understand the conversation. Some of the stuff being published right now is awful. But I also think we’re spending an enormous amount of energy on a question that, within a few years, is going to seem increasingly beside the point.
AI is changing how words are produced in much the same way it is changing how code, images, research, analysis, presentations and a hundred other things are produced. We can celebrate parts of that. We can complain about parts of it. We should certainly think seriously about the implications.
But it’s here. This is how work is going to get done. And once this becomes more normal, I don’t think anybody is going to care very much who did the typing.
We are in the awkward part
Right now AI writing is often obvious because we’re still early. A lot of people type something like, “Write me a LinkedIn post about the importance of customer experience,” hit Enter, copy the result and publish it.
People will look back at some of the things they’re publishing today and cringe. They’re learning a completely new way of working and so a learning curve is inevitable. But the tools will get better, and more importantly, people will get better at using them.
They’ll get better at teaching AI how they think. They’ll give it more of their previous work. They’ll correct it, reject things, and develop systems that remember those corrections. Eventually the distinction between “Eric wrote this” and “Eric used AI to write this” will be pretty fuzzy.
And I’m increasingly convinced that it doesn’t matter.
Writing is thinking. Typing isn’t.
One of the strongest arguments against AI writing is that writing itself is a thinking process. I agree. For many people, putting an idea into words is how they figure out what they actually think. Writing forces you to make distinctions, exposes holes in an argument and makes ideas concrete.
There will always be people who should continue sitting down and writing every sentence themselves because that process is important to them. But I think we’re making a mistake when we confuse writing as thinking with physically producing sentences with a keyboard (or a Blackwing 602 pencil if you’re a true purist).
Those aren’t necessarily the same thing.
I’ve been writing professionally for decades. I’m also old enough that typing (on a typewriter) was an elective in high school, which I never took. So I never became a particularly elegant typist. I hunt and peck, albeit at what I like to think is an impressively high velocity. Over the last couple of years I’ve also developed some pretty serious problems with my wrists.
So being able to sit down, talk through an idea into a microphone and have technology turn that thinking into organized prose is not some sad replacement for writing. For me, it’s fantastic.
I can explain something the way I would explain it to another person. I can wander a little, double back, correct myself, introduce an example, then realize halfway through that the example is actually the point. Then I can have AI help organize all of that.
And yes, sometimes, or if I’m honest most times these days, it kicks my ass at the actual writing.
It can take ideas that took me twenty minutes to ramble through and organize them better than I would have after an hour of typing, deleting, cutting, pasting and hitting Backspace a bazillion times. Then I react to what it produces: No, that’s not quite what I meant. This paragraph should come first. You’re overstating that. That’s not how I talk. You missed the important part. Try this again.
That process is still thinking. In some ways, I’ve found that it’s more iterative than writing the old way because I can move so quickly between an idea, an articulation of the idea and my reaction to it.
And the systems I’ve built for myself keep learning from those corrections. Not just how I sound, but how I think.
That’s much more interesting.
AI didn’t invent slop
Bad content did not arrive with ChatGPT. That old blogging joke exists for a reason. Long before generative AI, companies were publishing thousands upon thousands of completely forgettable blog posts written by humans. “Five Ways to Improve Your Customer Service.” “Why Digital Transformation Matters.” “Three Trends Every Business Owner Needs to Know.”
A human being could spend four hours writing something and still have absolutely nothing to say.
AI’s contribution is not the invention of meaningless content. It’s just dramatically reduced the cost of producing it. That’s a real problem because the algorithms have spent years training all of us to feed them: post more, publish more, stay visible, keep the content calendar full.
Now we’ve handed everyone a machine capable of satisfying that appetite almost infinitely. So yes, we’re going to get a lot more junk.
But the problem isn’t that a machine wrote it.
The problem is that there was no idea underneath it.
That’s the dividing line
If your entire content process begins with, “Write me a post about AI strategy for advertising agencies,” I can pretty much guarantee the world does not need that post.
The model has to invent the observation, invent the argument, invent the examples and invent the point of view. All it can really do is generate a statistically plausible version of what people generally say about the subject.
That’s where slop comes from.
But give the same system a practitioner who’s spent twenty years working with agencies, who has noticed something strange happening inside client organizations, who disagrees with the conventional wisdom, who has a story from yesterday’s meeting that illustrates the point, and who can explain why everyone else is looking at the issue incorrectly, and now you’ve got something.
AI can help turn that into an article. It can organize it, challenge it, condense it, find repetition, help adapt it for LinkedIn and suggest where the argument loses momentum. Increasingly, it can also learn enough about the person doing the thinking that it stops flattening everything into Generic Internet Professional Voice.
None of that bothers me, because the valuable part was never the keystrokes.
The valuable part was the insight.
So I’m not terribly interested in detecting AI
Maybe this post has some detectable AI fingerprints in it. I don’t know, and I don’t particularly care.
If the idea is weak, spotting an AI phrase doesn’t make it weaker. If the idea is useful, an awkward AI phrase doesn’t suddenly make the idea worthless either. We’re going to get better at this, the models will get better at this, and the weird stylistic tells we’re all obsessing over today will change or disappear.
Somewhere along the way, I think we’ll stop asking, “Was this written with AI?” We’ll go back to asking the question we should probably have been asking all along:
Was it worth reading? Because whether you’re writing with a pen, pounding on a keyboard, dictating to a human assistant, working with a ghostwriter or sending tokens through a transformer, there is one part of the process technology can’t rescue.
You still have to have something to say.