Since burning through tens of billions of dollars on its flop “metaverse” concept and laying off thousands in the aftermath of that gamble, tech giant Meta has strained to reinvent itself as a company poised to capitalize on the overhyped AI revolution. Last year, for example, founder and CEO Mark Zuckerberg unveiled animated AI chatbots modeled as alter egos of celebrities including Snoop Dogg, MrBeast, Paris Hilton, and Kendall Jenner.
But licensing the voices and likenesses of famous people did little to endear Meta to the younger demographics it wants to turn into loyal users of Facebook and Instagram. It junked the bots — widely ridiculed as creepy, corny, and without real purpose — less than a year later. This move coincided with the debut of its AI Studio, where users can create their own chatbots, including wholly fictional characters or, in the case of popular creators who want followers to feel more connected, “an AI that can message with your audience on your behalf, mimicking your tone and expressions.”
Now, Meta is planning to take the next step: integrating these AI creations as Facebook and Instagram “users” in themselves. As reported by the Financial Times, the hope is that these semi-independent custom avatars will prove more engaging to the young people who are crucial to the survival of Meta’s flagship social networks. “We expect these AIs to actually, over time, exist on our platforms, kind of in the same way that accounts do,” Connor Hayes, Meta’s vice-president of product for generative AI, told FT. “They’ll have bios and profile pictures and be able to generate and share content powered by AI on the platform . . . that’s where we see all of this going.”
The prospect seems an unusual one for websites predicated on the idea of human-to-human interaction. While Meta’s other AI offerings, including photo-editing tools, a ChatGPT-like text bot, and forthcoming software for producing AI-generated video, have more readily obvious appeal, it’s far from certain that an Instagram or Facebook populated by virtual apparitions would prove enticing to newcomers or satisfying for long-time users. Meta revealed to FT that users have created “hundreds of thousands of characters” since AI Studio launched in July, but that most of them remain private. The company declined to comment further to Rolling Stone.
Whatever the reaction, such a change to the fundamental structure of Meta’s platforms could be transformative. Recent months have brought a wave of alarm about our growing attachments to AI, from student overreliance on it in the classroom and its ability to spread inappropriate content or dangerous misinformation to a recent lawsuit from a mother of a teen who claims his intimate involvement with a Character.AI bot led to his suicide. That company did not comment on the pending litigation but told CNN that it was “heartbroken by the tragic loss of one of our users.” (AI dating programs are a burgeoning business particularly fraught with emotional hazard.)
The envisioned change also comes at a time when Facebook in particular is dominated by AI-generated spam images commonly referred to as “slop.” These pictures range from visually attractive though nonexistent landscapes and houses to depictions of babies, soldiers, the American flag, and Jesus Christ, presumably meant to evoke a response from older and conservative-leaning audience. This low-effort engagement farming allows the individuals posting the images — many of them in the Global South — to monetize the pages or sell related merchandise. Through payments to content creators who develop large followings for slop, Meta effectively incentivizes the content.
Some of that engagement no doubt comes from spam bots, which are automated to like, share, and comment on posts in order to drive interactions. But the implications of a Facebook where the majority of the activity is bots bouncing off one another are hard to grasp, even at a moment when artificially gamed exchanges are common on the social network.
Behind all this, Meta has been working to streamline its path to AI leadership while steadily rolling out more AI features to its 3 billion monthly active users. In January, Zuckerberg announced the merger of its Fundamental AI Research (FAIR) division with its GenAI product team, saying the move would accelerate their work in this field and was an important step on the way to creating a so-called “artificial general intelligence.” An AGI is a hypothetical autonomous system capable of teaching itself and surpassing human intelligence.
Experts told Rolling Stone at the time that such a breakthrough is still rather far-fetched, while Meta’s efforts to leverage user data in order to engineer its AI tech posed significant privacy risks. They also questioned Zuckerberg’s promise of transparency in the process and goal of “responsibly” open-sourcing an AGI — making the code available to the public — noting that Meta has offered little insight into its existing models and there is no real framework for safely releasing the blueprint for an (as yet theoretical) AGI.
All told, then, the latest AI spin from Meta sounds like more of the same: a mix of vague hype, dubious business models, and a baffling sense of what people really want from a social media network. If, in a few years, Instagram and Facebook are just places for AI bots to hang out, it stands to reason that the humans may find other ways to communicate.