Fewer humans, more AI: Meta’s true turning point
Or how Meta plans to hypnotize you with mind-like AIs.
Meta has announced the end of human moderation on Facebook and Instagram. In its place? “Community notes” and algorithms. Officially, this move is framed as a push to protect freedom of expression, reduce censorship, and streamline policies on sensitive topics. But beneath this polished narrative lies a far more transparent goal: accelerating the flow of content and maximizing user engagement. After all, fewer safeguards inevitably lead to more virality—and more polarization.
But this is just the beginning. The real story lies in how this strategy paves the way for massive AI integration.
Zuckerberg’s company has already begun testing features like AI-generated versions of oneself on Instagram, allowing users to create personalized avatars. Chatbots based on specific individuals or fictional characters are also in the works, designed to “enhance social interaction” on the platform (or what’s left of it). Finally, AI-powered video editing tools promise to reinvent audiovisual content, making every post more immersive and engaging (regardless of what gets sacrificed along the way).
But perhaps the most fascinating (or terrifying) aspect is the vision that seems to be emerging. You see, Meta recently published a paper exploring a field that is both highly technical and deeply symbolic: the application of “Theory of Mind” to language models. This ability, unique to humans, involves understanding what others think, believe, or feel.
According to the study, even the most advanced AIs, like GPT-4 or Llama, fail at complex Theory of Mind tests, sometimes scoring close to zero. And that’s where it gets interesting: why would Meta invest in frameworks like ExploreToM to evaluate and train its AIs on these tasks? Certainly not out of philosophical curiosity.
If Meta is delving into these concepts, it’s because they’re building something. Something far greater than just a generative content tool. They’re working to equip their AIs with “social intelligence,” a simulated understanding convincing enough to captivate our minds. This isn’t about creating AI to understand humans—it’s about perfecting the imitation of understanding.
Now, let me plant another seed. Most LLMs are built using datasets of human-generated online content. Take X, for example, which has already announced plans to exploit its platform data—content that’s often more conversational, reflective, and instinctive. You get the picture.
And what about Meta? Officially, it’s all quiet. But think about it: Facebook, Instagram… years of deeply personal data. Not just public posts, but private messages, spontaneous confessions, family memories—everything that, over the years, tells the story of our lives.
Do you really think Meta is leaving this treasure trove untapped? A database so intimately tied to our vulnerabilities, a mirror so finely tuned to the human experience…
And why is Meta building so many data centers? Perhaps for something far more ambitious, as discussed with my friend
:Zuck becomes a god. Meta, the church. The dollar fades, replaced by a “Z” currency to rival Musk’s “X.” All orchestrated by AI-driven financial advisors.
Let me remind you of an exchange from the early days of Facebook that perfectly encapsulates this mindset. When a journalist incredulously asked how Mark Zuckerberg managed to convince thousands of Harvard students to give him their personal data, here’s how he responded:
Zuck: Yeah so if you ever need info about anyone at Harvard
Zuck: Just ask
Zuck: I have over 4,000 emails, pictures, addresses, SNS
What? How’d you manage that one?
Zuck: People just submitted it.
Zuck: I don’t know why.
Zuck: They “trust me”
Zuck: Dumb fucks.
Those words, crude as they are, resonate with chilling clarity today. If this is the foundation, imagine what’s being built on it now.
In 2023, leaked internal meetings revealed that Meta was considering acquiring Simon & Schuster, one of America’s largest publishing houses, to enrich its AI models with high-quality literary works. And imagine if Facebook were to purchase a company like Automattic, the owner of WordPress, which itself acquired Day One, a popular digital journaling app.
You’ve got it: Meta isn’t just creating tools—it’s crafting illusions. It’s becoming the next Copperfield of the digital age. By exploiting intimate human narratives, it’s likely aiming to create interactions so flawless they become hypnotic.
Its true turning point? Realizing that a mind is captivated when shown its own reflection, polished by algorithms and illuminated by its own flaws.
But when every interaction is optimized to maximize our attachment, can we still tell the difference between what we truly want and what the algorithms want us to desire?
So tomorrow, ask yourself this: will you still be the author of your story, or just a supporting character in the AI’s? Unless, in a final twist, there’s nothing left but bots talking to bots. Poetic justice, perhaps.
MD