AI Gets "Sloppy" On Social Media

Welcome to this week’s Deep-Fried Dive with Fry Guy! In these long-form articles, Fry Guy conducts in-depth analyses of cutting-edge artificial intelligence (AI) developments and developers. Today, Fry Guy dives into AI “slop” on social media. We hope you enjoy!

*Notice: We do not receive any monetary compensation from the people and projects we feature in the Sunday Deep-Fried Dives with Fry Guy. We explore these projects and developers solely to showcase interesting and cutting-edge AI developments and uses.*

AI-generated pictures and posts are taking over social media, from fake African plastic bottle art to pictures of extravagant (yet fabricated) waterparks in Greece.

Millions of people believe pictures like these are real. Unable to discern the difference between authentic images and AI-generated ones, people are becoming easily deceived. The crazy part? This is just the beginning. Modern generative AI is only two years old. This means fake pictures posted on platforms like Instagram and X are only going to get more realistic, and they are likely going to explode in number.

Why is this happening? And why are sites like Facebook doing little to stop it? What does the future of social media look like, given all this AI content? Let’s explore these questions.

Did you see the recent dog airplane? It’s super cool!

As you might have guessed, this picture is completely fake. In fact, upon critical examination, there isn’t even a window for the pilot to see out of! Nonetheless, hundreds of people online believed it was legitimate, shared it with their friends, and even asked how they could get tickets to ride on it.

This example is one of thousands that flood social media every hour. From intentionally deceptive political images to fake pictures of athletes, AI-generated images are emerging all over the place and are deceiving the public. Countless people comment on these types of images, saying things like, “This is incredible!” Or in the case of the dog airplane, “Amazing paint job!”

These AI-generated pics have gotten so bad that there’s even a new word for it. They’re now coined as AI slop.

Digital slop is spreading like wildfire on the internet and can be found on almost every webpage. But despite the name, this slop isn’t so sloppy. It’s getting harder and harder to tell if many of these pics are real or AI-generated. Sure, we can all make fun of our non-tech savvy relatives and friends all we want, and think they’re completely naive to believe these AI pics are real. But don’t laugh too hard, because AI is coming for you too. No matter how immune you think you are to AI-generated content, or even if you graduated from the Ivy League, in the coming years, it’s going to be almost impossible to tell the difference between what is real and what is created by AI. In fact, a recent Turing test presented participants with an image of a person and asked them to identify whether it was AI-generated or authentic. The results were shocking, as 87% of respondents mistook at least one AI-generated image for a real photo of a person and only 62% of respondents interested in AI and machine learning managed to answer more than half of the questions correctly. Another, similar study, found that when presented with five images, only 1.9% of 31,118 participants accurately identified all five correctly. This is as good as a random guess! And given that AI is still an infant, this task is only going to get more difficult.

If AI-generated content is getting more and more convincing, you’d think trillion dollar companies like Meta would be doing something about it right now. That way, authentic content could be protected from content intended to deceive the public. You’d think that they’d start clearly labeling images created with AI. But actually, they’re doing the exact opposite. In a recent update, Facebook made a shocking change on how they label AI-generated images. As of September 2024, Facebook removed their AI warning label from pictures generated by AI. Instead, they moved the warning to an obscure side menu that 99.9% of the public never see. X has followed suit, allowing for the creation of images directly on the platform via Grok. Without any watermarking technology or restrictions, people are easily creating images of Donald Trump, Kamala Harris, and other important figures in seconds. And many of these photos are tricking the public!

If this slop problem is getting worse, why in the world are social media platforms like Facebook loosening their AI labeling policies? It’s actually quite simple: AI-generated posts make these companies tons of money in ad revenue. They have little concern that AI posts are tricking the public because they’re getting rich off of them. So instead of attempting to stop it, they throw their arms up and blame someone else, or say that it’s the government’s problem.

At first blush, this may seem like another case of corporate greed, and maybe it is. But let’s be honest with ourselves. A post of a mac-n-cheese swimming pool at Buc-ee’s is much more fun to look at than a picture of someone’s cat playing with a new toy. Because these AI-generated posts get millions of views, it ultimately means more ad revenue for Zuckerberg and his friends. So, contrary to what these social media companies say to the public on how they’re going to “protect users from AI content,” they secretly love the proliferation of these deepfakes. It’s a cash cow for them, so how can you blame them?

Social media companies are accumulating so much traffic from this slop that many of them are now actively helping users create AI posts for themselves. According to Axios, Meta is planning on generating images for users based on their interests alone—images that are created out of thin air, but that users are encouraged to post as their own. Facebook is also adding an AI feature that can alter a user’s post to make it more interesting. In other words, Meta will leverage AI to enhance your boring posts or pictures and make them appear way more exciting. If you live in a run down part of the city but still want to flex on social media, just tell Facebook to turn that abandoned building into a tropical resort and to throw a Caribbean beach in the background. In seconds, all your friends will think you are living your best life.

To be clear, Facebook isn’t alone in helping users create AI content. TikTok, Snapchat, Instagram, X, and other social media platforms have similar tactics. Because of this, by the year 2026, experts project that 90% of all of social media content could be AI-generated. Just think about that for a second: in the very near future, every 9 out of 10 images you see on Instagram will be AI-generated. If this projection is correct, you might as well throw your phone in the toilet, because you probably won’t care to scroll through a bunch of slop.

Social media will eventually turn into a spam-filled wasteland that’s going to drive everyone away. Some people call this the “Dead Internet.” The social networks originally built to connect humans will become a feeding ground for AI bots. This will be similar to a site that exists right now, called Chirper.ai, which is literally a social network where “no humans are allowed.” It’s a place where automated AI bots rule and interact with one another.

There has to be a solution to all of this, right? Well, to be honest, the answer is that the internet and social media as we know it is probably screwed. But this doesn’t mean all hope is lost; it just means we need to rethink social media. As AI-generated slop floods the social media feeds we once found ourselves addicted to, we will soon grow bored with pictures of mac-n-cheese pools and crave authentic, human content. To meet this demand, it’s likely human-only social networks will emerge. These networks will have strict verification processes to confirm human identity, similar to banks. On top of that, these same networks might mandate that users only make posts with certain anti-AI phones or software.

This is just one thought, and there will probably be many more as this narrative plays out. But one thing is for certain: anti-AI rhetoric and the urge for human-created content is going to intensify in the imminent future. People are soon going to be just as disgusted by digital slop as they are by food slop.

So if you want to be a tech hero, start working on the next great social media site—one that cleans up all of the AI slop. It’s going to be a messy job, but if you’re successful, you might become a billionaire in the process. Look out, Zuckerberg!