The anger is real, which is why it is worth taking apart rather than either endorsing or diagnosing it. Seven complaints sit under the word, and they are not equally strong.
1. It moves work onto you
This is the best complaint, and the one the people who coined the term led with. Something was generated in seconds and sent, and now you have to decide whether it is usable. The sender saved a few minutes; you spend much longer.
Cory Doctorow's framing, reported in a Columbia University convening on the topic, is that sending unreviewed machine output to strangers amounts to coercing them into unpaid labor, because the one thing the sender did not do is evaluate their own material. Simon Willison's original argument was the same idea stated more politely: sharing unreviewed generated content with other people is rude.
This complaint has the advantage of being checkable. You can measure the effort ratio, and the workplace research has: the 2025 BetterUp and Stanford study of 1,150 US desk workers puts the average cost of a single workslop incident at 1 hour 56 minutes.
2. It takes work away from people who do it
The economic complaint, and it is documented rather than hypothetical.
Wikipedia's documentation of AI slop records the case of a chainsaw carver whose sculpture style was reproduced in synthetic images posted by accounts claiming the work as their own. His statement is quoted directly: that this is a huge issue for carvers all over the world who are missing out on the credit and exposure their work should have earned. The same article documents creators whose likenesses, styles and audiences have been captured by synthetic accounts.
For illustrators, voice actors, translators and photographers, the complaint has two parts, and only the first gets discussed. The first is losing jobs. The second is losing attribution, which for a working artist is a pipeline rather than a trophy.
3. It makes everything look the same
Aesthetic flattening. Style guides, sentence rhythms, page layouts and image lighting converging on the same defaults, because everyone is prompting the same models.
The ACM paper on why slop matters describes the mechanism as overtly idealized output that flattens individual variation into canonical patterns. Anyone who reads for a living has noticed it. The writer Wesley Yang's observation about em dashes is quoted in academic writing on the topic as evidence of how recognizable the pattern became.
This is a real cost and a hard one to act on, because homogeneity has no single author to complain to.
4. It poisons the well you drink from
Trust erosion, and the most consequential complaint of the group.
Consider the documented case of fabricated images attached to a real disaster. During the response to Hurricane Helene, synthetic images circulated as evidence of events that did not happen, including one of a displaced child clutching a puppy. A Columbia University report describes a similar dynamic at the political level: a manipulated photograph posted without any label, in an environment where a year of cartoonish generated content gave officials room to dismiss it.
The mechanism the report identifies is worse than any single fake. The volume of obviously synthetic material normalizes the category, and deliberate manipulation hides inside the noise. One researcher describes slop as functioning as camouflage rather than as the payload.
5. It degrades the places you use to find things
Search results, marketplaces, streaming catalogues. The complaint here is about a channel becoming useless rather than about any particular item being bad.
This is where most people encounter the problem first, and it explains why the word spread through general audiences so quickly. A search result page full of pages written to rank is a different kind of failure from a funny generated video, and it is the one that costs people hours.
Research framing of the term's normative status makes a related argument about where the real harm sits: complaining that a particular output is bad is the weak version of the complaint, because quality depends on purpose and standard. The stronger version is about the aggregate restructuring the attention environment and raising the cost of finding anything worth reading.
6. It is insulting, in a specific way
The social complaint, and the reason the word has maintained its bite.
Being sent generated content without review communicates something about how the sender regards your time. That is the reading legal scholars give the phenomenon when they describe the resentment it generates, and it is why the offence scales downward so oddly: a long, grammatically perfect document can feel more insulting than a short careless one.
7. It arrived wearing a costume
The last complaint is not really about slop, and it is worth naming so it can be argued about honestly.
A philosopher writing in npj Artificial Intelligence argues that the word lets two very different claims travel under one label. In an optimist's mouth, calling something slop quarantines the failure: this is not what the technology is for. In a critic's mouth, the same word says the failure reveals what the technology actually is. Neither speaker has to state which claim they are making.
That is why conversations about slop so often end up being conversations about whether AI is good, without anyone having said so. Both readings are positions a person can hold. Only one of them is a claim about the document in front of you.
The case against the backlash
Any honest account has to include the arguments on the other side, and there are two good ones.
Slop answers a real demand. Researchers writing in the ACM make the case that dismissing slop as digital pollution misses its function: people want more content than human producers can supply, and generated material meets that demand at a price the market will pay. That is an argument about a system, not a defense of any individual item, and it is hard to refute from a position of taste.
Some of it is genuinely good, or at least interesting. The same paper argues that slop has aesthetic value, comparing it to other low cultural forms that critics dismissed before later taking them seriously. Plenty of people enjoy AI-generated video, find the absurdity funny, and would not trade it for more prestige television. Telling them they are wrong is not an argument.
There was always slop. Formulaic films, empty committee reports, bad novels and templated journalism all predate every model in existence. Critics on both sides make this point, and it cuts against the idea that the phenomenon is new rather than newly cheap.
Where the argument actually stands
Strip it down and the defensible complaint is narrow and specific: content produced without care, published without review, delivered at volume, and paid for by whoever reads it. That complaint is about effort and accountability, and it applies to a human-authored rush job just as well.
The complaints that fail are the ones that use the word to settle a larger question. Whether the technology is worth having, whether its aesthetic is a decline or a new form, whether people who use it are lazy: none of those are answered by looking at the thing in front of you, and all of them get smuggled in when nobody says which claim they are making.