Get Ready for a Lot More Bots on Social Media (2024)

screen time

By John Herrman, a tech columnist at Intelligencer Formerly, he was a reporter and critic at the New York Times and co-editor of The Awl.

Last month, a new social-media app started making the rounds. On SocialAI, the pitch goes, you’re always the “main character,” you’ll always get engagement, your replies will always be full, and you’ll have exactly the types of followers you want. The catch is that none of these followers are people —they’re all characters generated by AI.

It’s the sort of joke-shaped start-up that seems to exist solely to make a point. Its creator suggested as much, emphasizing the “thesis” of his app rather than why people might actually want to use it: that there are “tons of use cases that a broadcast model of LLM interaction” — social media — “has to offer that a chat interface simply cannot.” Maybe that’s true, but not on SocialAI, not now. In practice, posting for a crowd of unconvincing AI characters is about as boring and uncomfortable as it sounds. The app’s founder, former Facebook employee Michael Sayman, said that he had big plans to improve the app and seemed to take the predictably mixed feedback — “this is maybe the most embarrassing thing I’ve ever seen?” — in stride, perhaps because a lot of people in the industry were cheering him on.

Get Ready for a Lot More Bots on Social Media (3)

Photo: Screenshot, Social AI

This week, he shared an update: “I’m excited to share that after the success of SocialAI, I’ll be joining Meta to help build the future of Generative AI as the next chapter of this journey,” he posted on human social media. “Building SocialAI and seeing the impact the AI social network had across the tech industry made it clear that there was a natural alignment with Meta in shaping the future of AI.”

What does Meta, a company built on wringing money out of massive amounts of actual human attention, see in a project like this? Sayman says he “can’t share every detail” but is looking forward to “pushing the boundaries of what’s possible with Generative AI,” and to “create the tools that shape the future of how we connect and interact online.” What could that mean? A Facebook with more AI and fewer people? An Instagram stocked with sycophantic bots? A WhatsApp for talking with AI agents?

A fully automated Facebook, à la SocialAI, is of course absurd —at the very least, a social network without content created by people is no longer a social network. A substantially automated Facebook, however, is plausible, not least because a partially automated Facebook —a Facebook where users prompt content from AI, and where AI fills gaps in their feeds automatically— is already here. While a single hire at a 70,000-person company probably can’t tell us much, the state of Meta’s actual platforms and its recent product decisions can certainly tell us something: It is going to try it all.

In the broadest sense, modern social media is all about automation, in that users are induced to post by automated systems, their content is collected, sorted, distributed, and recommended mostly automatically, and advertising is sold, targeted, and displayed through a centralized software apparatus. There is still plenty of human participation and intervention involved in running Meta’s platforms —hence the 70,000 employees and tens of thousands of non-employee contractors behind the scenes —but it’s still helpful to understand modern social media in terms of human labor and its replacement, and competition between platforms as largely about new forms of automation.

Social media still depends on the creation, by people, of enormous amounts of content. If users don’t post, other users won’t have much to look at. Social platforms can automate a lot of processes, but they can’t easily automate the content itself. The business model has always faced a risk: What happens if users stop making as much good stuff? As a result, in recent years, social media companies have been approaching the question of what you might call user automation in both subtle and, more recently, fairly direct ways.

You might characterize the rise of Instagram as a story about streamlined photo editing or attribute some of Snapchat’s success to its (automated) face filters. TikTok has pushed this concept further, prompting users to participate in trends and making it extremely easy to produce videos. (The company is working on a feature in TikTok shop that will “generate scripts using artificial intelligence, so creators have enough to say and can sell more products during livestreams,” according to The Information.) Lend it your face and a bit of your personality, and TikTok’s video creation tools will streamline the creation of postable stuff. The platform has already succeeded at replacing feed-based social media, which drew on the users’ sharing and following activities to figure out what to show them next, with pure machine learning.

Meta, for its part, has long pursued similar strategies. The company’s recent and massive investment in generative AI has accelerated this tendency, trickling down to its social platforms in the form of features that are inching closer and closer to simply posting for users. The company has added its chatbot to Facebook, Instagram, WhatsApp, and Messenger, where it plays multiple roles. In WhatsApp and Messenger, it behaves a bit like another user on the platform, which has at least some conceptual overlap with SocialAI. On Instagram, it’s merged with the search function, but also prompts users to generate content —“generate a 1990s arcade” — for posting in their feeds. It’s also experimenting with customizable AI characters and suggesting users chat with them alongside their real friends. Bots, the scourge of social media, are now part of its product road map.

On Facebook, Meta’s oldest and shabbiest platform, the experimentation has been more extensive. AI is generating content summaries, showing up to answer questions in groups, and suggesting comment responses. In a September announcement, Meta described a new suite of tools that, used together, do most of the posting for you:

We’re expanding Meta AI’s Imagine features, so you can now imagine yourself as a superhero or anything else right in feed, Stories and your Facebook profile pictures. You can then easily share your AI-generated images so your friends can see, react to or mimic them. Meta AI can also suggest captions for your Stories on Facebook and Instagram.

Like SocialAI, these announcements read a bit like a punch line to a long and obvious joke: Who needs people on social media? In addition to nudging its users to start posting AI-generated content, including AI-generated images of themselves, Meta also said that it would start shoveling AI-generated content into users’ feeds itself, as a test. “[W]e’re testing new Meta AI-generated content in your Facebook and Instagram feeds, so you may see images from Meta AI created just for you (based on your interests or current trends),” the company said.

Get Ready for a Lot More Bots on Social Media (4)

Illustration: Facebook

It’s committed to the bit! It’s also really just pushing old concepts a bit further, exploring the limits of how much “help” its users want or will tolerate. A goofy TikTok face filter with AI-generated backgrounds and a clumsily promoted image generator on Facebook are features that share a category but are distinguished by execution and, perhaps, self-awareness. You know something has entered the water supply at Meta HQ when you start seeing posts like this:

Get Ready for a Lot More Bots on Social Media (5)

Illustration: Threads

Or maybe it’s extremely self-aware? Perhaps the wider conceptual argument behind the artificial-intelligence boom and/or bubble is that self-awareness is overrated and unnecessary.

The company could also just be responding, as it long has, to trends set by its users. Without Meta’s intentional help, Facebook has become the primary platform for the distribution and monetization of surreal, disgusting, and generally awful AI slop. It has been widely taken as a sign that the Facebook platform is irredeemably lost and that such slop is destined to take over the rest of the internet. Meta clearly disagrees, at least with the first point. By providing tools for its users to “easily share … AI-generated images” and by injecting AI-generated images that users didn’t ask for into their feeds, Meta is enthusiastically and all but explicitly getting into the slop business itself.

And why wouldn’t it? Facebook AI slop, as reported in this magazine, is a growing and lucrative business, aligned well both with Facebook’s general tendency to reward engaging content as its specific content monetization schemes circa 2024:

[AI slop] pages make money through Facebook’s Performance bonus program, which, per the social network’s description, “gives creators the opportunity to earn money” based on “the amount of reach, reactions, shares and comments” on their posts. It is, in effect, a slop subsidy. The AI images produced on Stevo’s pages — rococo pictures of Jesus; muscular police officers standing on the beach holding large Bibles; grotesquely armored gargantuan helicopters — are neither scams nor enticements nor even, as far as Facebook is concerned, junk. They are precisely what the company wants: highly engaging content.

Users are using off-platform AI to generate content to which other users seem to respond. For a social-media platform, the most natural next step in the world is to try to bring more of that action in-houseand to have it all unfold on the company’s own terms. Meta’s challenge here, as usual, is that it can’t quite go all the way —it can give users tools to generate infinite uncanny content, but it can’t give them tools to make Shrimp Jesus and Reverent Steroid Cop. Slop is engaging because it’s maximally strange, bad, weird, or cynical. It has, at least compared to the examples Meta is putting out itself, a disguised, malign, but distinctly recognizable human touch.

Still, even against this grim backdrop, SocialAI feels extreme, like a cautionary tale about what happens after you pass the tipping point from augmented social media (engagement heaven) to automated social media (empty unmonetizable hell). As its founder suggests, SocialAI is probably best understood as a different way to interact with AI rather than a model for social media. From certain perspectives, though —like that of, say, a social-media executive who seems quite fed up with friction and criticism from real people —it also represents a serene fantasy: a platform on which everyone does exactly what you want, infinitely and forever.

Tags:

  • john herrman
  • screen time
  • technology
  • meta
  • facebook
  • instagram
  • artificial intelligence
  • More

Show Leave a Comment

Why Meta Is Filling Your Feeds With Bots
Get Ready for a Lot More Bots on Social Media (2024)
Top Articles
Latest Posts
Recommended Articles
Article information

Author: Msgr. Benton Quitzon

Last Updated:

Views: 5470

Rating: 4.2 / 5 (43 voted)

Reviews: 82% of readers found this page helpful

Author information

Name: Msgr. Benton Quitzon

Birthday: 2001-08-13

Address: 96487 Kris Cliff, Teresiafurt, WI 95201

Phone: +9418513585781

Job: Senior Designer

Hobby: Calligraphy, Rowing, Vacation, Geocaching, Web surfing, Electronics, Electronics

Introduction: My name is Msgr. Benton Quitzon, I am a comfortable, charming, thankful, happy, adventurous, handsome, precious person who loves writing and wants to share my knowledge and understanding with you.