Tech experts hope new term for carelessly automated AI webpages and images can illuminate its damaging impact
Your email inbox is full of spam. Your letterbox is full of junk mail. Now, your web browser has its own affliction: slop.
“Slop” is what you get when you shove artificial intelligence-generated material up on the web for anyone to view.
Unlike a chatbot, the slop isn’t interactive, and is rarely intended to actually answer readers’ questions or serve their needs.
Instead, it functions mostly to create the appearance of human-made content, benefit from advertising revenue and steer search engine attention towards other sites.
Just like spam, almost no one wants to view slop, but the economics of the internet lead to its creation anyway. AI models make it trivial to automatically generate vast quantities of text or images, providing an answer to any imaginable search query, uploading endless shareable landscapes and inspirational stories, and creating an army of supportive comments. If just a handful of users land on the site, reshare the meme or click through the adverts hosted, the cost of its creation pays off.
But like spam, its overall effect is negative: the lost time and effort of users who now have to wade through slop to find the content they’re actually seeking far outweighs the profit to the slop creator.
“I think having a name for this is really important, because it gives people a concise way to talk about the problem,” says the developer Simon Willison, one of the early proponents of the term “slop”.
“Before the term ‘spam’ entered general use it wasn’t necessarily clear to everyone that unwanted marketing messages were a bad way to behave. I’m hoping ‘slop’ has the same impact – it can make it clear to people that generating and publishing unreviewed AI-generated content is bad behaviour.”
Slop is most obviously harmful when it is just plain wrong. Willison pointed to an AI-generated Microsoft Travel article that listed the “Ottawa food bank” as a must-see attraction in the Canadian capital as a perfect example of the problem. Occasionally, a piece of slop is so useless that it goes viral in its own right, like the careers advice article that earnestly explains the punchline to a decades-old newspaper comic: “they pay me in woims”.
“While the precise meaning of ‘They Pay Me in Woims’ remains ambiguous, various interpretations have emerged, ranging from a playful comment on work-life balance to a deeper exploration of our perceived reality,” the slop begins.
AI-generated books have become a problem too. A prominent example came when amateur mushroom pickers were recently warned to avoid foraging books sold on Amazon that appeared to have been written by chatbots and contained dangerous advice for anyone hoping to discern a lethal fungus from an edible one.
Image-generated slop has also blossomed on Facebook, as images of Jesus Christ with prawns for limbs, children in plastic bottle-cars, fake dream homes and improbably old women claiming to have baked their 122nd birthday cake garner thousands of shares.
Jason Koebler of the tech news site 404 Media believes the trend represents what he calls the “zombie internet”. The rise of slop, he says, has turned the social network into a space where “a mix of bots, humans and accounts that were once humans but aren’t any more mix together to form a disastrous website where there is little social connection at all.”
Nick Clegg, the president of global affairs at Facebook’s parent company, Meta, wrote in February that the social network is training its systems to identify AI-made content. “As the difference between human and synthetic content gets blurred, people want to know where the boundary lies,” he wrote.
The problem has begun to worry the social media industry’s main revenue source: the advertising agencies who pay to place ads next to content. Farhad Divecha, the managing director of UK-based digital marketing agency AccuraCast, says he is now encountering cases where users are mistakenly flagging ads as AI-made slop when they are not.
“We have seen instances where people have commented that an advert was AI-generated rubbish when it was not,” he says, adding that it could become a problem for the social media industry if consumers “start to feel they are being served rubbish all the time”.
Tackling spam in inboxes required an enormous cross-industry effort and led to a fundamental change in the nature of email. Big webmail providers like Gmail aggressively monitor their own platforms to crack down on spammers and are increasingly suspicious of emails arriving from untrusted email servers. They also apply complex, largely undocumented, AI systems to try to detect spam directly, in a constant cat-and-mouse game with the spammers themselves.
For slop, the future is less rosy: the world’s largest companies have gone from gamekeeper to poacher. Last week, Google announced an ambitious plan to add AI-made answers to the top of some search results, with US-based users the first to experience a full rollout of the “AI Overviews” feature. It will include links as well, but users who want to limit the response to just a selection of links to other websites will be able to find them – by clicking through to “web” on the search engine, demoted to sit beside “images” and “maps” on the list of options.
“We’ve added this after hearing from some that there are times when they’d prefer to just see links to webpages in their search results,” wrote Danny Sullivan, the company’s search liaison.
Google says the AI overviews have strong safety guardrails. Elsewhere on the web though, slop is spreading.
