What is it?
The metaphor ‘AI slop’ is used to describe content made using generative AI that people see as low quality, made with little effort and produced in huge quantities. This can refer to images, videos, texts and increasingly software code. The word ‘slop’ has been in use since the 14th century, referring successively to dung or slime, liquid waste or pig swill.
What does this metaphor suggest about AI?
The metaphor treats AI as a kind of machine that produces waste. The stuff it produces is careless and messy. It also spreads and is hard to contain. People who use the word often add more liquid imagery: slop ‘floods’ feeds or ‘oozes’ into search results.
The pig-feed meaning adds a second image. Pigs are fed leftovers, and if AI systems are trained on content that other AI systems have already produced, the waste goes round in a circle. This echoes the way mad cow disease (BSE) spread when animals were fed on animal remains. Here the worry is that human knowledge becomes contaminated by its own synthetic leftovers and eventually collapses.
Where does the metaphor commonly appear?
The metaphor appears everywhere that people encounter machine-made content that they have not asked for:
Social media and search: Users complain that most of their Facebook or LinkedIn feeds are just full of AI slop. Some add ‘-ai’ to their searches to filter it out.
News and commentary: Headlines and news columns use it to produce quick fake images, videos and stories.
Work and education: ‘Workslop’ describes AI-made reports and emails that look finished but leave colleagues to fix them. ‘Academic slop’ and ‘podslop’ show the word moving into study and podcasting.
Politics: A newer set of terms, such as ‘slopaganda’, describes political messaging made with AI.
Dictionaries and research: Dictionaries, linguists and researchers track the word, and academic papers are written about the topic.
What does the metaphor highlight?
This metaphor highlights several aspects of (generative) AI:
Volume: The main complaint is not that one piece is poor but that there is so much of it and cheaply made.
Lack of effort and care: Academic definitions describe slop as something that takes far less effort to make than to deal with. Silbey and Hartzog add that slop puts the burden on the person who receives it, and Willison’s comparison with spam makes the same point.
Feeling of loss and contamination: The sense that the internet, search results and workplaces are being polluted, and that it is getting harder to find what is real and trustworthy.
What does the metaphor hide?
Using ‘AI slop’ and its many creative compounds and derivations as insults can hide some important aspects of AI:
Producers: By treating slop as a flood that simply arrives, slop draws away attention from the companies whose tools and platforms make it cheap to produce and profitable to spread, although coinages such as ‘slopper barons’ show that some users do point at the tech leaders behind the flow.
Variety: Kommers et al. have argued that slop meets a real demand for content that humans could not supply, and that it deserves serious study. ‘Slop’ can make all AI-made content sound equally bad and to wave away any AI-made uses, including useful or playful ones.
A longer history: AI speeds up and scales up an old problem, but the metaphor can make it sound new.
Whose interests does this metaphor tend to serve?
The phrase mostly serves ordinary users, critics and creators who want a name for something they dislike or are worried about. It helps those who want to defend human work. It can be used to separate harmful AI output from creative human authorship. It is also useful for people who want to sell ‘human-made’ or verified content, because it makes the unverified kind sound like waste.
