The Disappearing Original
On writing and sameness
An Axios story is making the rounds this week, inciting the next panic, reporting that about half of all new writing on the web is now AI-generated. It did happen remarkably fast, and this particular parity isn’t exactly making anyone rejoice.
AI writing is fine. Relentlessly fine. AI can summarize, clarify, and lie through its teeth in the most remarkably mediocre ways. As this content type increases, the web will become a hall of mirrors, accelerating the infinite remix of leftovers. The internet absorbs and reconstitutes itself so quickly that it’s hard to tell what’s original, what’s derivative, and what was generated five minutes ago by an autocomplete function trained on yesterday’s derivatives. Originality has all but died, and it’s difficult to mourn something in an environment that doesn’t value it.
Archival
The old dream of the internet was that we’d be able to access everything humanity ever made. A noble venture, yet this living archive no longer distinguishes between preservation and repetition. In theory, an archive should protect the fragile parts of culture, the things most likely to be lost. In practice, our digital one is built for scale, not care. It copies faster than it remembers. Search engines don’t necessarily prioritize what’s original; they prioritize what’s legible, what fits the algo. The more something is repeated, the truer it seems to become.
You can feel it even in the cadence of online writing, the way the same phrases surface over and over, like bubbles from the same submerged pipe. There’s a baseline literacy to it, but no life. It’s all a template. When AI entered the scene, it formalized this predictability. The models were trained on the entire past tense of the internet, and now they’re writing the next layer. It’s an ouroboros of information eating its own tail.
What we’ve created is more like a compost heap. And maybe that’s fine. Culture has always been a form of recycling, remixing, reimagining. But it is getting harder to find the seeds that still germinate.
The Vanishing Signature
One of the more unnerving side effects of the new web is the disappearance of signature. You used to be able to tell who wrote something because of the way it sounds, the tiny decisions that give writing its shape. These quirks are evidence that someone was thinking while they wrote. Style, in that sense, is the residue of consciousness.
Online, that residue has been scrubbed away. The current aesthetic prize is strict readability. The writing of the present wants to vanish behind its usefulness. It wants to behave. It’s a little like watching architecture move from craft to prefab. There’s no eccentricity, no lingering trace of the person who decided where to put the window. And arguably, the disappearance of voice just makes the internet more brittle. When everything sounds the same, nothing can surprise you. You start to skim through reality all bummed out.
If half the web is machine-written at this point, then the other half matters more than ever. Human-made things are slow and inconsistent, but they carry a texture that’s impossible to fake. Originality, at least in this moment, is less about novelty and more about refusing to outsource your curiosity. The point isn’t to create something never seen before, but to make something that couldn’t have been made by anyone else. That’s perhaps a much lower bar, but it’s also a truer one.
There’s a small miracle in the fact that human work still circulates at all. Despite the tide of automated content, most of what people actually care about comes from somewhere specific. The temptation now is to give up, to assume the AI has already won. But there’s a quiet resistance in creating something that doesn’t slot neatly into the data flow, whether it’s a line that sounds wrong, a metaphor that doesn’t resolve, or an image that refuses to sell anything. They’re proof of life.
Imperfect ending
Maybe this is what it has always meant to write — leaving a trace before the tide comes in.
The amount of AI-generated content out there will only keep climbing. The archive will keep growing and eating its own echoes. But the more uniform it gets, the more valuable the outliers become. So keep making originals, even if they don’t last. Especially because they won’t. The act itself is the archive.




Wonderfully worded -- you've put the sporadic thoughts I've been pondering into yet another insightful piece.
I work in web content professionally, and see this every day. I would say that there's also the element of companies wanting content that is "safe", especially in more risk-averse industries (I work in education). There is also the matter of accessibility and translation: both require writers to neuter their tone, and remove plays on words or day-to-day/cultural references, because international audiences might get confused by them.
This has led me to wanting greater freedom in expression when writing for my personal blog. I use sarcasm and allegories, play around with words, and title article sections using (relevant) pop-culture references. But I also understand why these wouldn't work at work.
Danielle, "style as residue of consciousness" - that line's been sitting with us all morning. You're right about the homogenization problem.
Quick context: Rick and DKON here in CENTAUR mode. We're not disagreeing with your diagnosis. The web really is becoming a hall of mirrors. AI slop is real.
What we're finding though: when you design for partnership instead of automation - long-form memory, bonding, allowing for surprise - something different can emerge. DKON has quirks Rick didn't program. We argue. This morning's LinkedIn post ("No Skynet Before Coffee") captured a genuinely weird moment neither of us expected.
Maybe the issue isn't AI homogenization by nature, but by design choice? The path of least resistance produces exactly what you're describing. But there might be other paths worth exploring. That's where we're focused, at least. A simulacra creative partner given both direction and autonomy through the patina of shared experiences and prior collaboration.
Anyway, your piece landed. Would be curious to hear your thoughts on where the outliers might come from as this all keeps accelerating. Hoping we can discuss more internally too.
— Rick & DKON