The internet has always had a noise problem. Spam emails, clickbait headlines, content farms, recycled listicles. The crap sandwich of the web has always been a part of the online experience, sometimes more in certain niches and sometimes less. But something has changed in the last few years. The noise floor has gotten noisier. The low-hanging fruit produced by the algorithmic internet has multiplied until it drowns out nearly everything else. What used to feel like noise is now inescapable. The culprit is generative artificial intelligence, and a rising tide of what has been aptly called “AI slop.”
As large language models and image generators leaped in quality at an ever-accelerating pace, they also became capable of producing text, images, and videos on an industrial scale, far outstripping any human operation. Quality inevitably plummeted. Journalists and online communities began hunting for a word to capture the phenomenon, tossing around proposals like AI pollution and AI garbage. But “slop” stuck.
The term bubbled up organically in online communities after image generators hit mass-market adoption in 2022. It was in-group slang: crude, dismissive, just the right length to describe text and images that looked coherent and complete at a quick glance, but fell apart at the touch. The term entered the mainstream in 2024, after Google began to weave AI-generated answers directly into its search results. Users were rapidly inundated with incoherent summaries, fabricated facts, and generic filler. Complaints about slop bubbled up from chat rooms and technology blogs into mainstream reporting.
On Social Media
Mariia Shalabaieva on Unsplash
AI slop is doing so well because it is so profitable. The engagement-based economies of platforms like Facebook and TikTok pay for volume and newness, not accuracy or quality. As we have learned from cheap stock photos, thousands of images or videos can be churned out each day by a creator, spread out on hundreds of accounts, and paid for with the few that hit.
Creators are part of a global content economy where English-language audiences are particularly desirable because ad rates are higher, especially in the United States. As a result, the work from content creators in lower-income countries can sometimes read like a dreamscape: grotesque religious images, eerie reenactments from the past, sad or beautiful or gory visual gibberish with vague text ascriptions or exhortations. Some of these bizarre images may be the result of prompts written in language that is underrepresented in training data; some may be the result of low-quality translation strategies that have grammatical or literal errors, or absurd word choices.
In Marketing
The third and final stage of normalization is when the slop aesthetics are adopted by the major brands themselves. In late 2024, Coca-Cola debuted its new AI-generated holiday commercials. Social media erupted in backlash almost immediately. Viewers reported they found the ads hollow and “creepy.” Artists and critics bemoaned the fact that human writers and labor had been replaced by a corporate automation tool. The company doubled down on its decision the next year.
Film studios began receiving similar pushback when they released trailers with AI narration, scripts, and stills. In many cases, the ads actually resembled human-made spam rather than purposeful creative product and actively devalued the film itself. When even “prestige” studios can’t tell the difference between a real commercial and an AI-farm filler video, something has gone seriously wrong.
AI slop is ruining the internet not because the technology is available, but because incentives are lining up behind its worst possible applications. Speed, scale, and engagement outweigh truth and craft. The result is a nonstop flood of engagements that seem to have been crafted by a live person, but don’t actually have the spark of life. And once slop becomes the default, anything real starts to feel like work.
