An experiment. Earlier today I wrote a short thought by hand, about a hundred and fifty words, the way I’d say it out loud. Then I gave it to an AI and asked it to turn that thought into a full blog post. This is what came back. Judge it however you like. That’s the point.
There’s a word doing a lot of work right now: slop.
AI-generated writing is slop. AI images are slop. AI music is slop. Human work, by contrast, has substance. It’s a tidy line, and it lets a lot of people feel safe on the right side of it.
I think it’s the losing side of the argument. Not because nothing made with AI is bad (plenty of it is), but because we have run this exact experiment many times before, and we already know how it ends.
We’ve been here before
Writing. In Plato’s Phaedrus, Socrates tells the story of an Egyptian king who is offered the invention of writing and turns it down. Writing, he says, will make people forgetful. They’ll have the appearance of wisdom without the reality of it, because the knowledge will live in marks on a surface instead of in their heads. We know this argument only because Plato wrote it down.
Printing. In 1492, the abbot Johannes Trithemius wrote In Praise of Scribes, arguing that monks should keep copying books by hand: handwork was devotional, parchment would outlast paper, and printed books were careless. He then had it printed, because that’s how you reached people.
Photography. When photography arrived, painters were told their craft was finished. The painter Paul Delaroche is said to have declared that “from today, painting is dead.” Baudelaire, reviewing the Paris Salon of 1859, attacked photography as a refuge for failed painters, a mechanical trick that should know its place as a humble servant of the real arts. Painting didn’t die. Freed from having to record reality, it went on to Impressionism and everything after.
Recorded music. In 1906, John Philip Sousa, the most famous bandleader in America, published an essay called The Menace of Mechanical Music. Phonographs and player pianos, he warned, would kill amateur music-making and drain the soul out of performance. He wasn’t entirely wrong about everything, but nobody today thinks recorded music is soulless by definition.
Drum machines. When the Roland TR-808 came out in 1980, it was a commercial flop. Musicians mocked it because it didn’t sound like real drums. Session players worried about their livelihoods, and in 1982 the UK Musicians’ Union debated a motion to ban drum machines and synthesizers from recording sessions. Within a few years, cheap second-hand 808s were the backbone of hip-hop, electro and house. Today it’s one of the most loved sounds in music, precisely because it doesn’t sound like real drums.
Photoshop. Version 1.0 shipped in 1990. Within a decade, “photoshopped” had become an insult meaning fake. Film photographers held that digital wasn’t real photography. Now nearly every photograph you have ever admired has been through a digital darkroom, and nobody asks.
Auto-Tune. Cher’s Believe made it famous in 1998, and critics spent the next fifteen years calling it the death of real singing. It’s now just another instrument in the rack.
The pattern
Every one of these follows the same arc:
- Fear. A new tool does something that used to require years of human skill.
- Dismissal. Its output is declared soulless, fake, lazy: slop, before we had the word.
- Moral framing. Using it becomes a kind of cheating. Real artists don’t.
- Quiet adoption. The best people in the field start using it anyway, and make things nobody could make before.
- Forgetting. It becomes ordinary. Nobody remembers they were against it.
Then the next tool arrives, and we start again at step one.
AI is somewhere between steps two and four right now. You can see it in the word itself. “Slop” doesn’t describe a quality; it describes an origin. It’s applied before anyone has looked closely at the thing. That’s the tell. When a judgement is made about where something came from, not about what it is, it’s not a judgement about substance at all.
Where substance actually comes from
Here’s what the history teaches, if you read it the right way round: the tool never decided whether the work had substance. The person using it did.
A camera didn’t make Ansel Adams. The 808 didn’t write Planet Rock. Photoshop has produced both the worst advertising of the century and some of its most striking images. The difference was never the machine. It was intent, taste, and the thousand small decisions about what to keep and what to throw away.
AI is the same, only faster. It can produce enormous amounts of mediocre material with no effort at all, and a lot of what floods the internet today is exactly that. But humans have been producing slop for as long as there have been printing presses and content farms. Keyword-stuffed SEO articles, stock photos of people laughing at salads, identical pop songs written to a formula: none of that needed a neural network.
Slop is what you get when nobody cares. That was true before AI, and it’s still true now.
What stays human, and I’d argue becomes more important, is everything around the generation:
- Wanting something. A prompt is a desire made specific. Knowing what you want to say is still the hard part.
- Taste. Recognising which of a hundred outputs is the good one, and why.
- Editing. Cutting, reordering, pushing back, starting again.
- Meaning. Deciding what the work is for, and standing behind it with your name.
A camera still needs someone who knows where to point it.
The fair objections
This isn’t a claim that there’s nothing to worry about. There are real questions about AI: whose work the models were trained on and whether they consented, what happens to people whose jobs are built on skills that suddenly become cheap, and how we find the good stuff when the volume of everything goes up a thousandfold.
Those are serious, and they deserve serious answers. But notice that they’re about how the technology is built and used, and who benefits. None of them is an argument that work made with AI can’t have substance. Conflating the two lets people skip the hard conversation and go straight to the sneer.
So, is this post slop?
I made an album this year that’s credited to an AI collective, via the conduit: me. One of its lines says it better than I can:
Artificial describes origin. Not consequence.
This post started as a hundred and fifty words I typed on my phone. An AI turned them into what you’ve just read, with the history, the structure and, you can decide, maybe a better argument than my original. I gave it the idea, the stance and the direction. It did the research, the expansion and the phrasing.
If you found something of substance here, where do you think it came from?
And if you didn’t, then fair enough. But I’d ask you to say what’s wrong with the argument, not with who typed it. That’s all I’m asking of anyone judging anything made with a machine. It’s the same courtesy we eventually extended to the camera, the drum machine and Photoshop, just a little sooner this time.