A sizable portion of the internet has devolved into an AI-contaminated wasteland. While an easy solution remains elusive, a browser extension called Slop Evader offers a glimpse at what the internet used to be only a few short years ago.
While always prone to innumerable hazards, the online ecosystem is degrading largely due to the misuse of generative artificial intelligence content. Increasingly referred to as AI slop, this digital pollution often takes the form of torrents of uncannily realistic, wholly fake images and videos. These have inundated some of the most commonly used social media platforms, while countless shell websites push untrustworthy articles and blog posts penned by AI programs. Often trained on uncompensated human labor, these sites routinely game popular search engines that now prioritize AI results. This allows unverified AI slop to rise to the top of many queries while simultaneously burying actually legitimate websites underneath a mountain of garbage links.
It wasn’t always like this. In fact, you can arguably trace the shift back to a single date: November 30, 2022. That’s the day OpenAI debuted ChatGPT to the public, likely forever changing how we interact with the online world. Barely three years since its rollout, and it’s already hard to remember the relative ease with which you could find answers to a search query—or even simply trust the images and videos displayed in front of you.
To help raise awareness to just how bad it’s gotten, environmental engineer and artist Tega Brain created Slop Evader. After installing the extension on Mozilla Firefox or Google Chrome, users can employ it to search pre-AI archives on a handful of websites including Reddit and YouTube.
“This sowing of mistrust in our relationship with media is a huge thing, a huge effect of this synthetic media moment we’re in,” Brain told 404 Media in an interview published last week. “I’ve been thinking about ways to refuse it, and the simplest, dumbest way to do that is to only search before 2022.”
The AI slop’s deteriorative effects may not only be affecting the internet itself. Earlier this year, an MIT Media Lab study suggested that large language model (LLM) products like essay writing assistants are actually rewiring users’ brains. The resultant “cognitive debts” may even include weakened neural connectivity and damage memory retention, as well as wider “long-term educational implications.”
Slop Evader’s capacities will likely remain extremely limited, not to mention it’s impossible to provide any up-to-date information from after 2022. Even still, it starkly illustrates generative AI’s disorienting effects on an internet-reliant society. If nothing else, you can be relatively confident that searching for recipes on Reddit via Slop Evader won’t result in something like glue-laced pizza sauce.
