In the early noughties, give or take a few years when Mark Zuckerberg was writing algorithms allowing users to rank women based on their looks, Real Madrid under Florentino Pérez was putting together a galaxy of footballing superstars: Luís Figo, Zinedine Zidane, Ronaldo Nazário, David Beckham, and Roberto Carlos. They were called the Galácticos – bought for huge fees from clubs across the world. Zuckerberg now appears to be trying to replicate that to create the next big thing in artificial intelligence: Artificial General Intelligence, a literal deus ex machina.
In June 2025, Zuckerberg launched Meta Superintelligence Labs (MSL), poaching the world’s top AI researchers from OpenAI, DeepMind, Anthropic, and beyond.
AI was once something that only bothered prophets in the wilderness, but the launch of ChatGPT in 2022 brought generative AI into the lives of regular folks who never bothered about it before. ChatGPT, for those who don’t understand anything about AI, is basically a generative narrow AI language model trained to understand and produce text – not even close to a baby as far as cognitive abilities are concerned.
But it has changed how we think, how we communicate, how we use punctuation marks (hello em dash), and made managers of all ilk think they can simply replace humans with a GPT. At its core, it’s like a million monkeys with typewriters, and probably a tad less intelligent.
But AGI, or Artificial General Intelligence, is something completely different.
Strip away Silicon Valley’s Messiah Complex and AGI – Artificial General Intelligence – is simply an AI that thinks like us, learns like us, reasons like us, and then some. Today’s AI, like ChatGPT, can draft your resignation email in passive-aggressive corporate speak or compose a sonnet about samosas. But AGI is a different beast altogether.
Imagine an AGI that writes a Theory of Everything before breakfast, designs better AI chips to upgrade itself by lunch, cures cancer in the afternoon, and composes a Mahler symphony to unwind by evening. It doesn’t just perform tasks it was trained on; it generalises intelligence across domains like us – except at terrifying speed and scale.
Why does Zuckerberg want it? Because whoever builds AGI first doesn’t just win the AI race. They redefine civilisation itself. An AGI could run Meta’s platforms autonomously, invent new technologies that keep it ahead forever, replace thousands of employees overnight, and transform Zuckerberg from a social media baron into a techno-philosopher king presiding over the most powerful mind humanity has ever created.
So, Zuckerberg launched Meta Superintelligence Labs and poached 11 AI titans from OpenAI, DeepMind, Anthropic, and beyond. From Lucas Beyer, co-author of Vision Transformer, to Johan Schalkwyk, the speech AI legend behind Google Voice Search, it’s a roll call of AI royalty offered $50–100 million packages to build AGI.
But here lies the AGI paradox. Once it arrives, it won’t remain politely at human level. It will improve itself into superintelligence, seeing us the way we see ants: with indifferent curiosity or casual genocide.
Meta Superintelligence Labs is his final moonshot. But as history shows, the Galácticos didn’t win much. Only time will tell if Mark’s Galácticos can take the giant leap. Of course one is exaggerating a bit but to borrow a line from my father: “Khayali pulav mein ghee ka kami kyun? (Why be stingy with ghee even in your daydreams?)”