David Silver Leaves DeepMind: The Creator of AlphaGo Is Going After Superintelligence

The AI world has just lost one of its most stable constants. David Silver, a man whose name at DeepMind meant no less than that of Demis Hassabis, has announced his departure. If you remember the legendary 2016 match in which AlphaGo crushed Lee Sedol, know that Silver was behind it. He did not just write code: for decades he argued that reinforcement learning is the only right path to real intelligence. And now that path is taking him away from Google.

Why is this happening right now? Context matters more here than dry facts. Lately DeepMind has turned from a cozy lab for geniuses into a huge corporate factory, merged with Google Brain. When organizations get too big, they slow down. Silver, who always preferred elegant algorithmic solutions to endlessly "grinding" data on servers, clearly felt cramped. His exit is a clear signal: today the most ambitious problems are solved not in open-plan offices of thousands of people, but in compact groups of fanatics.

Silver's goal sounds modest: superintelligence. But unlike the marketing promises of most Silicon Valley startups, Silver has something to build on. His work on AlphaZero and MuZero showed that systems can learn without human involvement, inventing strategies we never dreamed of. His new project will probably be built around the idea of "pure" learning, in which AI does not merely imitate human text but learns the logic of the world by interacting with it. That is a fundamentally different approach from the one we see in today's chatbots.

This exodus recalls the recent moves of Ilya Sutskever, who left OpenAI for his own project, SSI. We are watching the fragmentation of AI's "old guard." The people who laid the industry's foundations ten to fifteen years ago are leaving their gilded cages one by one. They take with them not only knowledge but also a reputation that will draw the best talent and billions in venture investment. Google will survive, of course, but without Silver its research division risks losing the very spark of mad innovation that made it a leader.

What does this mean for us? Most likely, a new player is coming, one that will not play by the rules of "ship the model faster so the stock goes up." Silver has always played the long game. If he has decided it is time to step out of Google's shadow, the puzzle of what the next qualitative leap in technology should look like has probably already come together in his head. We are entering an era in which individual geniuses again matter more than the computing power of cloud giants.

Bottom line: the era of DeepMind's dominance in fundamental research has officially ended. Can Silver repeat AlphaGo's success on his own, or will superintelligence prove a tough nut to crack even for him?