Mark Zuckerberg wants to feed you more AI slop

By Parmy Olson

Have you started to notice more AI-generated nonsense on social media? All over Facebook feeds there are bizarre images of Jesus made from shrimp, suspiciously perfect log-cabin interiors and photorealistic children holding professional-looking paintings. And Mark Zuckerberg doesn’t just expect this flood of machine-made content to continue — he’s working to accelerate it.

Unlike the captains of other tech giants who are building AI models and then selling access to them, the chief executive of Meta Platforms Inc. has made his free. Meta’s Llama 3.2 can generate text and images and is marketed as open source, meaning anyone can modify its code. Software developers have used Llama to make apps that turn personal photos into works of art or to create “personally-tailored” marketing content.

According to Zuckerberg, these efforts aim to stop a consolidation of control in AI. “What you want to prevent is one organization from getting way more advanced and powerful than everyone else,” he said in February. But he offered additional reasoning on Meta’s third-quarter earnings call this week: Free AI tools mean more AI slop, a potential boon for his platform.

When an analyst on the call asked about the proliferation of AI apps, the Meta CEO talked about how developers could use Llama to create more AI content that “just makes people’s feed experiences better.” Here’s his full comment, with my own emphasis:

“If you look at the big trends in Feeds over the history of the company, it started off as friends, right? So all the updates that were in there were basically from your friends posting things. And then we went into this era where we added in creator content too, where now a very large percent of the content on Instagram and Facebook is not from your friends…

“And I think we’re going to add a whole new category of content, which is AI generated or AI summarized content or kind of existing content pulled together by AI in some way. And I think that that’s going to be just very exciting for the -- for Facebook and Instagram and maybe Threads or other kind of Feed experiences over time.”

Get ready, folks. In much the same way that short videos and viral content took over feeds once populated with posts from our friends and family, the next wave of content will be machine-generated. A progression from personal to viral content and now to AI content seems like a dystopian direction for a social media firm that’s long framed itself as “connecting people.” But Zuckerberg calls this new trend “promising.”

His view is not unusual in the industry. I’ve spoken to several technology executives who believe that AI-generated content — which could make up as much as 90% of content on the Internet, according to one wild estimate — will be accepted as the new normal. AI-generated videos will eventually be called “videos,” the thinking goes.

Facebook has already been flooded with bizarre imagery dubbed AI slop. Earlier this year, researchers who investigated more than 100 Facebook pages posting high volumes of AI content found that some were posting dozens of these images each day. Some were content creators trying to grow their followers, while others were scammers trying to advertise non-existent products or drive traffic to ad-filled websites.

The slop can have a real-world impact. On Thursday, thousands of people in Dublin, Ireland were reportedly duped into visiting a non-existent Halloween parade after seeing AI-generated ads that spread across social media, according to disinformation researcher Ciaran O’Connor.

Content farms once flooded Facebook with low-quality, human-written articles to game the algorithm, but AI tools let you generate far more content at no cost. In effect, it’s a new form of clickbait, and researchers have shown that Facebook’s algorithm happily promotes it.

Which brings us to the real reason Zuckerberg made his AI free. By getting AI models into the hands of as many developers as possible, Meta is getting more AI-generated content onto the web. The higher the volume of AI content, the more likely there will be posts that can go viral on Meta’s platforms to keep people scrolling. That’s critical for a company that generates 98% of its revenue from advertising.

It’s also a shrewd strategy that’s straight out of the Silicon Valley playbook, according to Julien Cornebise, an associate professor at University College London:

Consider how in the mid 2010s, Google campaigned for the proliferation of fiber broadband, investing billions in fiber infrastructure that had nothing to do with the company’s core business. At the time, Google positioned itself as advancing internet access for the public good, but there was an obvious strategic benefit: increasing the addressable market for Google ads and services. “More people online at high speed meant more people using Google,” Cornebise points out.

Zuckerberg seems to be following the same pattern, spending billions on infrastructure for others to freely use, and positioning that work as democratizing AI. In reality, he’s growing the pipes that will fuel his operation for years to come.

Let’s hope the public pushes back on the slop. Viral AI content might seem like harmless entertainment, but flooding people’s feeds with machine-generated noise crowds out the experience of genuine human expression and authenticity. Zuckerberg risks repeating an ignoble chapter in Facebook’s history when its algorithms rewarded clickbait farms and mass-produced content. He can hardly call it “social” anymore. Soon it won’t be “media” either.