If you only want the definition: AI slop is low-quality content produced with generative AI in bulk, published with little or no human review. That is the sense the dictionaries settled on, and it covers the way the word is used most of the time.
The interesting part is everything the definition leaves out, because everyday usage has drifted further than the dictionary has.
This page is about the phrase rather than the concept: how AI slop behaves in a sentence, which words it appears beside, and where the dictionary sense and the everyday sense diverge. For the concept itself, start with what is AI slop.
The dictionary sense
Merriam-Webster, which made “slop” its 2025 Word of the Year, defines it as “digital content of low quality that is produced usually in quantity by means of artificial intelligence.” Three ingredients: low quality, high volume, machine generation.
Cambridge Dictionary, which entered the term as a headword earlier, is tighter and slightly different: digital content such as images and video, of low quality, created by artificial intelligence. Its entry carries the labels informal and disapproving, which tells you more about how the word works than the definition does.
Google's own machine-learning glossary offers a definition from the producing side: output from a generative AI system that favors quantity over quality, with a web page full of cheaply produced low-quality content as the example. It is a small thing, but worth noting that the company most often accused of shipping slop defines the term in a glossary it maintains itself.
The everyday sense
Usage in the wild leans on a different hinge. The most-cited formulation comes from Simon Willison's May 2024 post, which argued that slop names content that is “mindlessly generated and thrust upon someone who didn't ask for it.” Note what is missing: any test of volume, and any test of quality. Under that reading, what makes something slop is that it was generated without care and delivered to someone who now has to deal with it.
Put the two side by side and the difference is a difference in emphasis:
| Dictionary sense | Everyday sense | |
|---|---|---|
| What is being judged | The content | The act of sending it |
| Volume required | Yes, usually | No |
| Quality required | Yes, low | Not necessarily |
| Typical complaint | “This is worthless” | “Why am I reading this” |
In practice most complaints about slop are making both charges at once, which is part of why the arguments get heated. Two people can agree that something is slop and disagree completely about which test it failed.
How it behaves in a sentence
Three usage notes that catch people out:
It is a mass noun. You say “that article is AI slop”, “the feed is full of slop”, or “a piece of slop”. Saying “an AI slop” sounds wrong for the same reason “an email spam” does: the word names a category, in the way spam and content do.
It usually takes no article in headlines. “Slop is taking over search” reads naturally. The definite article appears where a specific flood is meant: “the slop on my feed”.
It attaches to a platform or a medium. People write about slop on Facebook, slop in search results, slop books on retail listings. The word is more often used about a channel than about a named work, which is another sign that it is really a complaint about a system.
The compounds it has already produced
A word that catches on this fast starts breeding, and slop has bred quickly.
- Workslop. Low-quality AI output inside an organisation. A study by Harvard Business Review with Stanford and BetterUp coined the term for content that looks good but lacks substance, and found it landing on colleagues to check.
- Slopper. A person, usually used as an insult, who leans on generative tools to produce work they would otherwise have done themselves.
- Microslop. A nickname used by critics for Microsoft's AI-branded features and the output they generate, part of a broader pattern of attaching the word to a company or product name.
- Trendslop. Coined in the same academic line of work, for a tendency to reach for whatever is currently fashionable over what is actually reasoned.
- AI fatigue. The neighbouring term for the exhaustion people describe as its consequence: sorting real material from the rest.
What the word is really doing
Read enough uses and one feature stands out. Slop is a word people reach for when the relationship between effort and obligation has broken down. Somebody spent two minutes generating something and now you are expected to spend twenty minutes deciding whether it matters.
That is why the term lands hardest in workplaces and in search results, and why it feels excessive when applied to a single video someone made for fun. The offence is not the machine generation. It is the imposition.
It is also why the word carries a second, larger argument inside it. Writing in npj Artificial Intelligence, a philosopher makes the case that “slop” quietly switches between two verdicts: one on a particular output, one on the whole technology. In an optimist's mouth, calling something slop protects AI's promise by quarantining the failure. In a critic's mouth, the same word says the failure is the technology showing its true nature. Neither speaker has to say which claim they are making, and that ambiguity is the term's real power.
Using the word precisely
If you want your usage to hold up in an argument, add the test you are applying:
- “That is slop because nobody checked it before it shipped” is a claim about review.
- “That is slop because it says nothing specific” is a claim about substance.
- “That is slop because there are forty thousand of it” is a claim about scale.
Each of those is defensible and checkable. Leaving the test unstated is what turns a judgment about a document into a fight about AI in general, which is the failure mode the term invites and the one worth avoiding.