From Bizarre Videos to Fake Papers: AI Turning the Internet into a Massive Trash Heap}
AI-generated content, from creepy videos to fake research papers, is flooding the internet, creating a digital trash landscape and raising concerns about quality and authenticity.
Don’t let AI become a “garbage factory”.
Who would have thought that an AI-generated video could garner 252 million views on Instagram, with 3.257 million likes?
In the video, a woman in a bikini clumsily steps onto a boat’s railing and leaps into the sea, splashing water everywhere. The boat also loses balance and tips over due to her weight and movement.
Sharp-eyed viewers also noticed many AI anomalies, such as a man watching from the boat turning his head 360 degrees like an owl, and other illogical tilts of the boat.
This exaggerated scene aims to entertain but also raises concerns about body-shaming. Due to its viral success, the creator has produced many similar videos.
Further exploration of this creator’s other works reveals a series of disturbing images: pig-faced divers, zombie-like interviewees, and bizarre creatures—each more unsettling than the last. (AI horror content is intense, so videos are omitted 😂)
Today, almost every social media platform features strange, creepy AI videos. Recently popular is an AI-generated “cannibalism” video.
 |  |
The images depict food items like lemons, oranges, steamed buns, broccoli, and blueberry muffins, all with human faces, smiling and “eating” each other with spoons.
From Instagram blogger reallyweirdai
Creating these videos is surprisingly simple: just a prompt, and Veo3 can generate a surreal scene like an orange with a human face slowly biting into a slice, with detailed textures and a dreamy atmosphere.
If you find these videos somewhat cute, the next example is truly disturbing.
This “You Are What You Eat” video went viral on Reddit, showing food-based characters devouring their own kind—like a sushi person wolfing down sushi or a yogurt person gulping strawberries. The addition of sound makes it even more creepy.
This video, created by blogger Bennett Waisbren inspired by mukbangs, uses a dark aesthetic to highlight modern gluttony. While profound in message, it’s visually disturbing.
Reactions are mixed: some call it “creative” and “easy money,” while others find it “disgusting” and “nightmarish.”
This discomfort stems from the “uncanny valley” effect, where objects that look almost human but not quite evoke feelings of disgust and fear.
The food faces display realistic expressions—smiling, chewing—that trigger unease when mimicking human behavior.
Such bizarre AI videos flood social media, driven by naked traffic motives. Algorithms favor eye-catching, provocative content, which fuels curiosity and engagement, encouraging creators to produce more extreme and unconventional videos.
Low-cost AI tools make it easy for creators to generate highly realistic videos with minimal effort, further amplifying this trend.
More seriously, AI-generated “garbage” isn’t limited to entertainment—it's infiltrating academia too.
Researchers at Sweden’s Brose University found hundreds of AI-generated fake papers in Google Scholar, eroding trust in scientific literature and spreading pseudoscience at low cost.
Another study examined peer review, noting that words like “meticulous” (34× more frequent), “commendable” (10×), and “intricate” (11×) are favored by models like ChatGPT, increasingly used to assist or even replace human reviewers, especially near deadlines.
Some AI-produced papers are blatantly fake, such as a 2023 medical journal article with a cartoon mouse with reproductive organs, full of errors and nonsensical text, yet it passed peer review and was published, exposing serious issues.
Scientists’ reputations depend heavily on publication records, but this system is exploited by low-quality AI-generated articles, threatening the integrity of scientific knowledge and fairness.
Whether it’s creepy videos or fake academic papers, the harm and influence of AI go far beyond mere creation.
As AI continues to develop rapidly, we must ensure that the internet doesn’t turn into a vast dump of misinformation and low-quality content.
References: