In 2016, he developed an AI program at Google DeepMind, AlphaGo, that taught itself to play the famously difficult game of Go with a kind of mastery that went far beyond imitation.
Silver has since founded his own company, Ineffable Intelligence, which aims to create more general forms of AI superintelligence. Silver says the company will do this by focusing on reinforcement learning, which involves AI models learning new capabilities through trial and error. The goal is to create “superlearners” who surpass human intelligence in many areas.
This approach contrasts with how most AI companies plan to create superintelligence by exploiting the coding and research capabilities of large-language models.
Silver, speaking to WIRED from his office in London, says he thinks this approach will fail. As amazing as LLMs are, they learn from human intelligence rather than creating their own.
“Human data is like a kind of fossil fuel that has provided a wonderful shortcut,” says Silver. “You can think of systems that learn on their own as renewable fuels – something that can just learn and learn and learn forever, without any limits,” he says.
I’ve met Silver several times and—despite this proclamation—he always strikes me as one of the more polite people in AI. Sometimes, when talking about ideas he considers foolish, he smiles wryly. However, right now, he is extremely serious.
“I think of my mission as making the first contact with superintelligence,” he says. “By superintelligence, I mean something really incredible. It should invent new forms of science or technology or government or economics for itself.”
Five years ago such a mission would have seemed ridiculous. But tech CEOs now regularly talk about machines surpassing human intelligence and replacing entire categories of workers. The idea that some new technological twist could unlock superhuman AI capabilities has recently given rise to billion-dollar startups.
Ineffable Intelligence has so far raised $1.1 billion in seed funding at a valuation of $5.1 billion – a huge sum by European AI standards. Silver has also recruited top AI researchers from Google DeepMind and other leading labs to join his effort.
Silver says he will donate all the money he earns from the equity in Afable Intelligence to charity – a sum that could reach the billions if successful.
“Building a company focused on superintelligence is a huge responsibility,” he tells me. “I think this is something that should be done for the benefit of humanity, and any money I make from Infallible Will will go to high-impact charities that will save as many lives as possible.”
total focus
Silver met Google DeepMind CEO Demis Hassabis at a chess tournament when they were children, and the pair later became lifelong friends and collaborators.
They remained close even after Silver left Google DeepMind, which he did only because he wanted to chart an entirely new path. “I think it’s really important that there is a specific AI lab that really focuses a hundred percent on this approach,” he says. “This is not just a corner of some other space dedicated to LLM.”
Silver says that the limitations of the LLM-based approach can be seen with a simple thought experiment. Imagine going back in time and releasing a major language model into a world that believed the world was flat. He says that without being able to interact with the real world, the system will remain an enthusiastic flat-earther, even as it continues to improve its own code.
However, an AI system that can learn about the world on its own can make scientific discoveries of its own.
<a href