In 2016, the AI program he developed at Google DeepMind, AlphaGo, taught itself how to play the famously difficult game of Go with a kind of mastery that went beyond imitation.
Silver has since founded his own company, Ineffable Intelligence, which aims to build more general forms of artificial superintelligence. Silver says the company will do this by focusing on reinforcement learning, which involves AI models that learn new capabilities through trial and error. The vision is to create “super learners” who exceed human intelligence in many areas.
This approach contrasts with the way most AI companies plan to build superintelligence, by exploiting the programming and searching capabilities of large language models.
Silver, speaking to WIRED from his office in London, says he believes this approach will fail. As great as MBA students are, they learn from human intelligence, rather than building their own.
“Human data is like a kind of fossil fuel that provides an amazing shortcut,” Silver says. “You can think of self-learning systems as a renewable fuel, something that can learn and learn and learn forever, without limits,” he says.
I’ve met Silver several times, and despite that statement, he’s always struck me as one of the most humble people in the AI field. Sometimes, when he talks about ideas he considers ridiculous, he gives an evil smile. Right now, though, he’s dead serious.
“I think our mission is to make first contact with superintelligence,” he says. “By superintelligence, I mean something truly incredible. It must discover for itself new forms of science, technology, government, or economics.”
Five years ago, such a task might have seemed ridiculous. But technology CEOs now routinely talk about machines surpassing human intelligence and replacing entire classes of workers. The idea that some new technical development might unleash superhuman AI capabilities has recently given rise to a slew of multi-billion-dollar startups.
So far, Ineffable Intelligence has raised $1.1 billion in seed funding at a $5.1 billion valuation, a massive amount by European AI standards. Silver has also recruited top AI researchers from Google DeepMind and other leading labs to join his endeavors.
Silver says he will donate all the money he makes from Ineffable Intelligence stock – a sum that could reach billions if he succeeds – to charity.
“It’s a big responsibility to build a company focused on superintelligence,” he told me. “I believe this is something that needs to be done for the good of humanity, and any money I make from Ineffable will go to high-impact charities that save as many lives as possible.”
Total focus
Silver met Demis Hassabis, CEO of Google DeepMind, at a chess tournament when they were young, and the pair later became lifelong friends and collaborators.
They remained close after Silver left Google DeepMind, which he only did because he wanted to chart an entirely new course. “I feel like it’s really important to have an elite AI lab that’s really focused 100 percent on this approach,” he says. “It’s not just a corner somewhere else reserved for LLM holders.”
The limits of the MBA-based approach can be seen through a simple thought experiment, Silver says. Imagine going back in time and launching a big language model into a world that thought the world was flat. Without the ability to interact with the real world, he says, the system will remain flat-Earther-hungry, even if it continues to improve its code.
However, an AI system that can learn about the world on its own can make its own scientific discoveries.
