A few days ago I stumbled upon a website for a design and marketing agency in my area that I was genuinely impressed by. At first glance, they had all the hallmarks of a premium shop: Clean site, nonchalant copywriting, and a sharp portfolio featuring the kind of work that makes you sit up a little and think “okay, these people know what they’re doing.” So I did what any nosy creative professional does when he’s impressed by a stranger’s website: I went and read their blog just to see how they think.
Lo and behold, upon getting there, the first entry I found on the blog was a piece from the agency CEO, doing the one thing everybody and their mom is doing for easy eyeballs: arguing against AI-assisted content. Fine — that’s a position, and people are allowed to have it. Except the argument itself was doing something funny: it was vague about what the actual downsides of AI writing were supposed to be, while in the very same article admitting that AI-assisted blogging had already been documented as building web traffic for competing organizations. So: bad for SEO, bad for marketing, bad for content creation — allegedly — but also, somehow working? In so many words, the stance amounted to “we don’t fully know what the consequences might be,” which like, yeah, no shit: none of us do.
But that’s not actually what stopped me. What stopped me was the writing. There was no flow between the sentences. The vocabulary was thin. Each “section” of this piece masquerading as thought leadership was, at best, three or four sentences deep.
I sat there thinking: this is your platform? You’re the CEO of a company, you’re out here telling people AI content is dangerous, and the human-made alternative you’re personally producing is — generously — mediocre. It’s not well executed, it’s not well argued, and the world — or hell, even my personal perspective — isn’t meaningfully improved. So what, exactly, are we protecting here?
That’s the thought that cracked this piece open for me: if you’re good at making content, it doesn’t matter if you use AI. If you’re bad at it, however, it doesn’t matter if you don’t. The tool was never the differentiator. It simply got blamed as if it was.
That brings us to the actual question sitting underneath all of this, the one that’s going to define much of the cultural conversation for the remainder of 2026 and beyond: what is slop?
So What Is Slop, Actually?
Here’s the definition I’ve landed on after chewing on this for a while: slop is the byproduct of production processes executed with little to no reference for the essential characteristics that make a deliverable useful or valuable. Not “AI-made.” Not “cheap.” Reference-less. Slop is what happens when nobody bothers to ask what would actually make something good before they make it.
Conventionally we think about this in terms of media, and for good reason — media is the one category where slop achieves maximum deployment velocity while demanding minimum baseline effort. In the AI era specifically, you already know the examples, because your feed is full of them:
Blog posts masquerading as intellectual thought pieces that are nothing more than superficial subject-matter overviews wrapped in a boilerplate literary voice. E-books cobbled together from recycled quotes and bullet points, functioning as nothing more than bold-faced Shopify cash grabs. LinkedIn broetry-maxxers pumping your feed full of garbage takes, formatted in all-caps, double-spaced text that reads like it was written for a fourth grader who just discovered the return key.
And because the enshittification of content genuinely feels like it’s well underway in the form of junk emails, poorly wrought blog posts and repugnant, unoriginal e-book slop, the easy, correlative conclusion is that AI did this to us. But anyone who’s seriously considered the zeitgeist, though, knows that this explanation is an oversimplification at best. And it’s an oversimplification that only remains true if you believe in an illusion of what the full scope of human creativity has actually produced.
The Golden Age Fallacy
We love to celebrate the excellence of human creativity — fine art masterpieces, original written works, masterfully composed music, popular films, and yes, even design. But that instinct is also, quietly, a vaunted conception of what creativity intrinsically entails. And here’s the controversial, unvarnished truth underneath it: most expressions of human creativity are pedestrian, if not downright bad.
Think about it honestly. For every genuinely quality piece of media you consume, there are easily ten more in that same category that are poorly composed, poorly written, poorly ideated, poorly performed, poorly designed, poorly produced, or poorly delivered — and that’s being generous. I’m even including quality-but-undiscovered artists in that 1:10 skill-to-slop ratio, because the ratio isn’t about fame, it’s about execution.
All of this is said to say: the version of creativity we’re so desperate to preserve is, a lot of the time, predicated on an elitist notion of what creativity entails. The broader reality is that when all humans practice creativity, most of what gets made is shit. There are plainly more shitty paintings than there are masterpieces. There are more shitty books than there are NY Times best sellers. There are more shitty songs than there are Grammy winners, or hell, even just halfway decent hits that make it to the radio. There just are. And here’s the thing — we wouldn’t celebrate creative excellence at all if the opposite were true. Excellence is only meaningful because it’s rare.
To be blunt about the point: for as much credit as AI gets for being the factory of slop deliverables, it’s easy to forget that humans have been manufacturing slop content for a lot longer than AI has existed. There have always been D-tier movies. There have always been genuinely forgettable books, and musical “passion projects” so bad that even passion couldn’t justify their existence. The tabloids of the world — the Globes, the Enquirers — have always existed, always making bold, objectively false claims about the apocalypse, killer diseases, and secret love children between public figures and extraterrestrials. And unlike satire, these tomes of American slop culture never presented themselves as a joke. They wanted you to take them seriously.
AI didn’t invent slop. Slop is, more often than not, the byproduct of humans trying and failing to produce something — or, hell, sometimes not trying at all. What AI actually did was make the creation, ideation, and deployment of slop faster. And like every democratizing technology before it, it lowered the barrier of entry for the average person trying to make slop in the first place. But the people creating the slop are, for all intents and purposes, the same people who have always been creating slop — and were always going to create slop. Whether or not AI adoption is salient has nothing to do with it. Slop was always going to exist in some form.
The Interchangeability Test
Here’s where I think the current treatment of AI slop actually goes wrong. The argument against it, made by the people who support human-made content, tends to have an unintended side effect: it treats AI slop as interchangeable with quality human content. I don’t know who needs to hear this, but it’s not.
If AI slop and quality human content were actually interchangeable, we — as consumers — wouldn’t be able to reliably tell the difference between them. But that’s not the world we live in. We live in a world where, even with AI slop everywhere, plenty of people can still identify when a work is poorly executed, poorly performed, or blatantly derivative. Ultimately, this raises a bigger, more uncomfortable question: if AI slop could genuinely be swapped in for quality content with zero negative consequence to the end user, would it even be accurate to keep calling it slop?
I’d wager the answer is no. And that’s the paradox nobody arguing this seems to have considered: you can’t simultaneously treat something as an existential threat to human creativity and as garbage. Those two framings cancel each other out. If the output is genuinely bad — if it’s actually slop — then by definition it cannot out-compete good human work, because there’s already enough signal baked into it for people to correctly sort it as inferior. The existential-threat framing only works if the content is secretly good. The garbage framing only works if it isn’t. Pick one: You don’t get both.
Don’t get me wrong: there will always be people who write off all AI content — including the genuinely good stuff — as slop. Those people are incorrect in their assessment, full stop. Because regardless of how much bad AI content currently dominates the sphere, there are still examples of good AI content that people find objectively enjoyable, despite being clearly, obviously, blatantly AI.
The Lego Videos
A great example of this are the Donald Trump Lego videos being released by the propaganda wing of the Iranian government. I’m not a fan of Trump. I think the Iranian government is reprehensible. Neither of these facts are the point, and if you fixate on the politics of who made the thing and why, you’ll miss the actual lesson.
Here’s what actually happened: despite being obviously, unmistakably AI-generated, these videos have been watched by millions of people, who genuinely enjoy and keep watching them. I know this firsthand, because I’ve had at least two different best friends personally sit me down to show me the latest ones. My dad likes these videos. My mom likes these videos. These are wildly different people, from wildly different corners of my life, none of them coordinating with each other, all independently deciding a Lego-and-satirical-pop-song mashup about Donald Trump was worth interrupting my day for.
Is there an argument that these videos are still slop? Sure — the music is obviously derivative, and the visual language clearly rides on existing IP. But it effectively combines familiar images and familiar music into something that stands out against the sea of run-of-the-mill political media, and — whether we personally choose to appreciate the merits or not — plenty of people genuinely enjoy the result. Here’s the reality: AI-assisted work, like anything else, can be good work. What determines whether it’s good are the points of reference used to inform it — the research, the process, the thoughts, the opinions, the actual voice of the person behind it. Strip those away and that very same work becomes two-dimensional: it becomes superficially opinionated, devoid of literary character, and just plain old fucking boring. Keep them intact and a work becomes colorful, informative, rich in perspective, and genuinely entertaining. And for anyone who thinks AI is structurally incapable of hitting that second bar — boy, do I have a surprise for you.
Not only can AI meet that standard in the hands of a skilled creative, but disciplines like writing have already shown it can meet that standard consistently. What actually dictates an AI-assisted work’s quality, most of the time, are its points of reference — and that’s exactly where skilled professionals and subject matter experts will consistently outstrip the irrational ambitions of an unindoctrinated one-shot prompt engineer.
Pulling back from Trump and Iran and Lego bricks for a second — the actual formula underneath those videos is one that could explain success in almost any content category: take something unexpected, pair it with something familiar, layer in cultural or political salience, and you’ve got a deliverable people actually want. The “why” behind the content — the political strategy — shapes what gets made, but it fails to explain why it worked. What explains why it worked is that whoever made it already knew the formula and knew how to manipulate it. I don’t think a layman could have made the Iranian Lego videos. I think a marketing expert could have. The tool makes a wide range of choices possible. Expertise is what makes specific choices good.
What Skill Actually Means
A lot of the prompt-engineer crowd is buying into a vision of AI that may never fully materialize: that AI is a conceivable replacement for skill. The error in that thinking comes from assuming skill is only about what a person does. In reality, it’s just as much about what a person knows.
To a layman, a writer jots words on a page, a designer opens Figma and does “designer-y things,” and a coder writes useful gobbledygook. Never mind the complexities of discovery, planning, audience awareness, or any of the other judgment calls that can only be made through actual qualitative experience. Without those things and more, a person is fundamentally incapable of meeting the full slate of production needs required to make something useful — or even just good.
Even with every tool humanity has ever built, you still can’t one-shot an app with a single-sentence prompt. Hell, you can’t one-shot it with a three-page prompt either, and the reason is simple: apps are fucking complicated. They need complementary feature sets, sensible user flows, and reasonable solutions for the edge cases that inevitably show up no matter how carefully you planned. Anyone who thinks they’re going to ship a market-ready app on their favorite vibe-coding platform while lacking a modicum of awareness of any of that, is fundamentally kidding themselves.
The truth of the matter is that the questionable economics of AI businesses fundamentally incentivize these companies to sell their tools as accessible to everyone. But anyone who actually knows better also knows that successfully building a digital product on a vibe-coding platform was never an intended outcome for laymen. It was almost always intended for experts — people who already had the taste and the judgment, and just needed the labor sped up. The same argument holds for every other AI creative tool, writing included, no matter how pedestrian writing might seem as a general practice: the best AI-assisted apps, designs, videos, and written work will keep coming from developers, designers, videographers, and writers who were already specialized before the tool ever showed up, because they’re the ones most predisposed to actually leveraging it.
With all of that in mind, insinuating that a work is bad simply because it uses AI is almost as preposterous as suggesting a book is bad because of the pen that wrote it. Are we really going to act like prose matters more than ideas, when the enshittification of prose has been underway for a lot longer than AI has been around — and is more plausibly the result of defunded education programs and loosening communication standards from increased social media use? Be so forreal.
AI has never been the problem. People’s use of AI has — and a huge amount of that use is less about force multiplication than it is about filling a gap in skill or knowledge that was there well before the tool showed up.
Say the Accurate Thing
So — what is slop? It’s not a result of improved technology. It’s the result of expertise being absent in the production process, and it has been around since long before anyone typed a prompt into anything. AI didn’t create the appetite for slop, and it isn’t going to be the thing that cures our appetite for it either. What it did was hand a fast, cheap production line to people who were going to make garbage anyways — and it also handed a genuinely powerful tool to the people who already knew what they were doing. This is exactly why some AI-assisted work is bad, and some of it is comprised of Lego videos about Donald Trump that my dad, unprompted, insisted I watch.
The dividing line was never the tool. It was never going to be. It’s whether the person on the other end of the prompt actually has something worth saying, and the judgment to know how to say it. Everything else is just noise dressed up as a controversy.