"AI slop" was named word of the year by Merriam-Webster, by Macquarie, and by the American Dialect Society. Three dictionaries looked at the internet and picked the same word for it. They weren't wrong.
We run an AI studio. We get called a slop factory roughly once a week. Here's the uncomfortable thing: the people saying it are describing something real, and most of the time they're describing our competitors accurately.
So rather than argue, let's agree, and then be precise about where the agreement stops.
Slop isn't "made with AI." Slop is volume without intent. Content generated because it could be, rather than because somebody wanted it to exist.
You know it when you see it. Video that's fluent and says nothing. The same six stock compositions with a different subject pasted in. Text that reads like it was written by someone who's never held an opinion. Thumbnails of things that never happened, attached to stories nobody told.
The tell is always the same: nothing was thrown away. Slop is everything the machine produced, shipped. There was no moment where a person looked at it and said no.
The problem was never the tool. It's that generating got free and deciding never did, so a lot of people stopped deciding.
Post anything made with AI and you'll meet a particular kind of anger. Not criticism. Contempt. "Slop." "Clanker." The comments are rarely about the work in front of them and they're almost never interested in a reply.
It'd be easy to write those people off, and most studios in this category do. I think that's a mistake, because the anger is usually pointing at something true.
For a lot of people, the value of creative work was never only the finished thing. It was the fifteen years spent learning to make it. The bad first films. The nights in an edit bay. When a machine produces something passable in nine seconds, it reads as an argument that those fifteen years were a waste, and that the person who spent them is now surplus. That isn't a technology opinion. That's a person watching their identity get repriced in public.
I have some standing to say this. I wrote, produced and directed a feature film. I know what that costs, and I know what it feels like to watch the ground move. So when an editor calls me a clanker lover, I don't think he's stupid. I think he's frightened, and the meanness is what fear sounds like when it's got nowhere useful to go.
He's also, on the main point, right. Most of what's flooding his feed is garbage. He just aimed at the wrong target.
Quality is the surface of it. Underneath is something larger and far more legitimate, and it isn't aesthetic at all. It's economic.
A very small number of companies are capturing an extraordinary amount of value and power, very quickly, using material the rest of us made. That's the serious version of the complaint. Who owns the models. Who owns the compute. Who gets paid, who gets consulted, and who just gets restructured. Consolidation.
I'm not going to pretend I'm standing outside that. I use these tools every day and this studio exists because they exist. I'm downstream of the same concentration everybody's angry about, and I benefit from it.
But look at who's actually in the room. The decisions people are furious about get made by a handful of people running a handful of companies with an enormous amount of capital. Sam Altman. Dario Amodei. Larry Ellison. Whether you think they're building the future or hollowing it out, that's where the power is going.
It isn't going to a studio in Austin.
That ain't me, babe. Go direct your guns somewhere else.
When the contempt lands on a working practitioner making a rap battle about the Fourth Amendment, it's found the reachable target rather than the responsible one. I'm in the comments. The people who'd actually have to answer for the concentration are not, and they never will be.
And if consolidation really is the worry, and I think it should be, then the fights worth having are about compensation for training data, licensing terms, labour agreements, provenance standards, and antitrust. Those are real levers with real leverage. Calling a filmmaker a clanker lover is not one of them. It just makes the person nearest you feel bad, which I suspect is sometimes the point, and which changes nothing about who ends up owning the future.
Here's the part that gets skipped. The argument assumes there was a fair system before this, and that AI is wrecking it. There wasn't. The studio system was itself built out of economic consolidation, and it has spent a century deciding which stories get made and which artists get to eat. Harvey Weinstein was not an aberration the industry caught. He was a man the industry protected for decades because he was profitable.
It's also consolidating right now, in public, faster than anything happening in AI video. Paramount Skydance agreed in February 2026 to acquire Warner Bros. Discovery, a deal WBD shareholders approved in April. If regulators clear it, one company controls Paramount, Warner Bros., HBO and CNN. Twelve state attorneys general sued to block it on antitrust grounds. The Writers Guild filed its own suit. The writers' union is in court trying to stop a studio merger, which tells you where the writers think the threat is.
And there's a detail almost too neat to be true. Paramount Skydance is run by David Ellison. His father, Larry Ellison, co-founded Oracle, one of the companies building the compute that AI runs on. One family, both ends of the pipe. That's what consolidation actually looks like, and it isn't a prompt.
So I'm not defending AI against a golden age. I'm a small studio in Texas using new tools to make things the old system would never have greenlit, in a business that pushed most artists to the fringe long before a model generated a single frame. Being small and quick is the entire point.
You can be angry about who owns the future. I am. Just check the address before you send the letter.
Here's the line. Generating isn't making. Generating got free. Making didn't.
People assume the answer is brutal filtering. Generate ten thousand things, keep the one good one. That's still slop, you've just hidden most of it. If you need to throw away almost everything, you didn't know what you wanted before you started.
We work it like a shoot. Ask how many takes Scorsese shoots of a single shot. Plenty. Nobody calls those takes waste, and nobody calls the cutting-room floor slop, because every one of them was a deliberate attempt at a thing somebody had already decided should exist. A crew doesn't roll on everything that moves and sort it out later. Somebody writes a script, somebody breaks it into shots, and the day is spent getting coverage of things that were already decided.
Same here, and the numbers bear it out: about 80% of what we generate gets used, and when a shot misses we reshoot it in the next few minutes rather than losing another day on location. That's not us being lucky with a slot machine. It's what happens when the thinking is finished before the machine starts.
Same here. The volume is real, and generating in quantity is genuinely part of it. But it's coverage against a plan, not a lottery ticket. The discipline lives upstream, in knowing what the thing is before you make it. That's the part slop skips, and it's the part that still costs a person.
Underneath that there is a script somebody wrote, a story somebody argued about, and a named human who signs their name to the result. When we make something synthetic, we say so, out loud, on the page. Anyone selling you synthetic media who won't show you the seams is worth exactly as much suspicion as you'd expect.
We named the company Talentless AI. We're not going to be precious about this.
Useful whether you're hiring a studio or just arguing in the comments. Five questions:
→ What was decided before they started? Ask for the script, the shot list, the plan. If generating came first and the thinking came after, it's slop no matter how much got filtered out.
→ Who wrote it? A human should have written the story before a model rendered it. If the script came out of the same box as the pixels, it will feel like it.
→ Is anyone's name on it? Slop is anonymous, because nobody wants credit for it.
→ Do they disclose? A studio confident in the work says which parts are synthetic. Hiding it tells you what they think of the work.
→ Would it survive losing the novelty? Take away "made with AI." Is it still worth four minutes? Most of it isn't. That's the test.
AI slop is low-effort synthetic content made in volume with no particular intent behind it: templated video, fluent text that says nothing, images generated because they could be rather than because anyone wanted them. The word was named Word of the Year for 2025 by Merriam-Webster and by Australia's Macquarie Dictionary, and the American Dialect Society followed in January 2026. It's a real phenomenon and the complaint about it is legitimate.
No, but most of it is, and pretending otherwise is how the people making slop defend themselves. The tool doesn't decide. A hundred generated clips with nobody directing them are slop. The same hundred clips shot as coverage against a script somebody wrote, cut by an editor with taste, is a film. The difference is whether anyone decided what the thing was before the machine started.
Because most of what they see is genuinely bad, and because the volume of it makes everything harder to find. Underneath that is something more serious: for a lot of people the value of creative work was never only the finished thing, it was the years spent learning to make it. Watching a machine approximate that in seconds reads as an argument that those years didn't matter. That's a real fear and it deserves a better answer than a demo reel.
Ask what got decided before anything was generated. Slop is what happens when a machine runs first and somebody looks for something usable afterwards. Real work is shot like coverage against a script, so most of what comes back is usable and the misses get reshot. Then ask whether a human wrote the story, whether anyone's named as the author, whether the synthetic parts are disclosed, and whether it'd still be worth watching if you didn't know how it was made.
It was not a meritocracy. The studio system was built out of economic consolidation and spent a century deciding which stories got made and which artists got paid, and it protected Harvey Weinstein for decades because he was profitable. It is also consolidating right now: Paramount Skydance agreed in February 2026 to acquire Warner Bros. Discovery, twelve state attorneys general sued to block it on antitrust grounds, and the Writers Guild filed its own suit. Defending AI does not require pretending the old arrangement was fair.
Yes, and that's the most serious version of the complaint, much more serious than the one about quality. A handful of firms are capturing enormous value very fast using material other people made. Steve Mudd's position is that he's downstream of that and benefits from it, and also that a small studio in Austin isn't where the power is accumulating. If concentration is the concern, the levers that matter are training data compensation, licensing terms, labour agreements, provenance standards and antitrust, rather than hostility aimed at individual practitioners.
We work hard not to, and the safeguard is upstream rather than downstream. We generate in volume, but the way a crew shoots coverage: against a script and a shot list, of things somebody already decided should exist. That's why most of what we make gets used. If a studio has to discard almost everything it generates, that isn't rigour, it's a sign nobody decided what the piece was before they started.
A rap battle about the Fourth Amendment that the sitting Mayor of Knox County, who is in it, called amazing. An educational channel at 93,700 subscribers. A UAP documentary series built from ninety-five primary sources across twenty-seven countries. None of it's defended by the fact that AI made it. It's defended by whether it's any good.
Fair. Bring us something you think can't be made well and we'll find out together.
Let's talk