The AI Slop Economy: 75,000 Songs a Day and the Age of Perfectly Average Content

There is a number floating around the music industry that is hard to sit with. Deezer, the streaming platform, reports that around 75,000 AI-generated songs are uploaded every day, which is 44 percent of all new uploads. That is not a prediction. That is the current daily rate.

Music is just the most measurable slice of a much bigger thing. Merriam-Webster made "slop" its Word of the Year for 2025, defining it as low-quality digital content mass-produced through AI. An audit of American newspapers found about 9 percent of newly published articles flagged as partially or fully AI-generated. Every day, millions of blog posts, product descriptions, images, reels, and social posts are generated in seconds and published without a human stopping to ask whether they should exist.

The striking thing about AI slop is not that it is bad. Bad content is easy to identify and reject. The striking thing is that most of it is fine. It is grammatically correct, professionally presented, reasonably informative, and almost completely forgettable. It clears every approval gate because there is nothing wrong enough with it to reject. A marketing manager reads a 1,500-word AI article, checks that the keywords are there and the structure is logical, and says "looks good." Next week another one appears. Eventually the company has hundreds of articles that are technically competent and interchangeable.

That is the paradox people keep missing. AI did not invent generic marketing. It industrialized it. The bottleneck was never production. The bottleneck was deciding whether a piece of content deserves to exist, and AI made it possible to skip that decision entirely while still meeting every output quota. The result is what one writer called the coming age of extremely average content: an enormous middle ground of material that is good enough to survive but not good enough to be remembered.

The consequences are showing up everywhere at once. Search engines have started penalizing unvetted mass publishing, which means the volume play that worked for a while is now a liability. Consumers report fatigue with over-polished content, and "having a human feel" has become a trust signal precisely because it is scarce. The People's Daily in China ran a piece telling the short-drama industry to bring the "human feel" back. The pushback is global because the flood is global.

The AI face problem in China's short-drama boom is the same phenomenon in visual form. When every AI-generated male lead has the same face, viewers feel a physiological disgust, not because the face is ugly but because it is interchangeable. The average face is the visual version of the average article: competent, smooth, and completely forgettable. Both come from the same root, which is a model optimized for the safest possible output run without a human to push it toward the specific.

The music numbers are the most concrete warning sign, because music is where the flood is easiest to count. 75,000 AI songs a day means the catalog grows by millions every month, most of it never played by a human listener. The platforms are caught between two pressures: they want the growth that AI content brings, and they do not want to be seen as warehouses for synthetic filler. Deezer and others have started filtering and labeling, but the scale of the problem is such that labeling feels like bailing out an ocean with a cup.

There is a subtlety to the slop problem that is easy to miss. It is not that AI produces bad work. It is that AI produces work that is good enough to get approved, and "good enough to get approved" is the enemy of "good enough to matter." A competent but forgettable article passes every gate because there is nothing wrong with it to reject. A slightly worse article with one embarrassing error would actually get caught. The slop survives precisely because it is mediocre in the most inoffensive way possible.

There is a useful way to think about what is actually scarce now. Content is infinite and getting cheaper. What is scarce is a reason for the content to exist. A specific experience, a real opinion, a number the writer actually looked up, a detail that would not survive a paraphrase. The companies and creators who thrive in this environment are the ones who treat AI as a first draft and spend their effort on the part only a person can do: deciding what is worth saying.

The economic logic has already flipped, and most people have not caught up. When production was expensive, the company with the most output won. When production is free, volume is table stakes and differentiation is the only lever left. A brand that publishes a hundred average articles competes with a million other brands publishing the same hundred average articles. The one that publishes one genuinely useful piece wins the attention, because usefulness is the one thing a probability model cannot fake.

There is a counterintuitive conclusion hiding in all of this. The rise of slop does not devalue human writing and judgment. It does the opposite. When everyone can generate a thousand average things, the ability to recognize and produce the one non-average thing becomes more valuable, not less. Editors, fact-checkers, and people with actual taste stop being a cost center and become the entire moat.

The platforms are starting to reflect this in how they rank and recommend. Search engines penalize unvetted mass publishing. Streaming services label synthetic music. The direction is toward a system that rewards the specific over the volume, because the volume has become worthless. That is a slow process, and it will not happen uniformly, but the incentives are already shifting underneath everyone who publishes.

The slop economy is not going to be regulated away, and it probably should not be. It is the natural byproduct of making production nearly free. What it does is raise the value of the opposite. Authenticity, specificity, and editorial judgment were always valuable; they just used to be table stakes. Now they are the entire game, because when anyone can make a thousand average things, the only way to be noticed is to make one thing that is not average at all. That is the real lesson of the slop era, and it applies to every person and company that publishes anything online.

There is a way to test whether you are part of the problem or part of the answer, and it is disarmingly simple. Before you publish anything, ask what specific thing this piece knows that a model could not have generated on its own. If the answer is nothing, you are adding to the slop. If the answer is a real experience, a real number, a real opinion, or a detail only someone who was actually there would know, you are adding the one thing the flood cannot replicate. The slop era is not the end of human content. It is the moment human content stopped being the default and became the differentiator, and the people who understand that are the ones who will still be worth reading, watching, and hearing a decade from now.