AlphaGo Creator Warns AI Is On the Wrong Path
▼ Summary
– David Silver, founder of Ineffable Intelligence, aims to build superintelligence using reinforcement learning, where AI learns through trial and error rather than from human data.
– Silver contrasts his approach with most AI companies, arguing that large-language models fail because they learn from human intelligence instead of building their own understanding.
– Ineffable Intelligence has raised $1.1 billion in seed funding at a $5.1 billion valuation, and Silver plans to donate any profits from his equity to charity.
– Silver believes reinforcement learning can create “superlearners” that surpass human intelligence in many domains, discovering new forms of science, technology, or governance.
– He illustrates the limits of LLMs with a thought experiment: a model released in a flat-earth world would remain an avid flat-earther, unable to make its own scientific discoveries.
In 2016, David Silver introduced the world to something unprecedented: an AI program called AlphaGo that taught itself to master the ancient, deeply complex game of Go with a skill level that transcended mere imitation. Developed at Google DeepMind, it offered a first glimpse of superintelligence in action.
Today, Silver has left DeepMind to found Ineffable Intelligence, a company with a singular mission: building more general forms of AI superintelligence through reinforcement learning. This technique lets AI models learn by trial and error, effectively creating what Silver calls “superlearners” ,systems capable of outperforming humans across countless domains. The goal is to move beyond the current industry obsession with large-language models (LLMs), which Silver believes will ultimately fail to deliver true superintelligence.
“Human data is like a kind of fossil fuel that has provided an amazing shortcut,” Silver told WIRED from his London office. “You can think of systems that learn for themselves as a renewable fuel,something that can just learn and learn and learn forever, without limit.”
Silver has always struck me as one of the more humble figures in AI, often flashing a puckish grin when discussing ideas he considers silly. But right now, he is deadly serious. “I think of our mission as making first contact with superintelligence,” he says. “By superintelligence I really mean something incredible. It should discover new forms of science or technology or government or economics for itself.”
Five years ago, such ambitions would have seemed far-fetched. Today, tech CEOs routinely predict machines outpacing human intelligence and replacing entire workforces. The promise of a technical breakthrough unlocking superhuman AI has spawned a wave of billion-dollar startups. Ineffable Intelligence has already raised $1.1 billion in seed funding at a valuation of $5.1 billion,an enormous sum by European AI standards. Silver has also recruited top researchers from Google DeepMind and other frontier labs.
Remarkably, Silver plans to donate all proceeds from his equity in Ineffable Intelligence,potentially billions if successful,to charity. “It’s a huge responsibility to build a company focusing on superintelligence,” he says. “I think this is something that has to be done for the benefit of humanity, and any money that I make from Ineffable will go to high-impact charities that save as many lives as possible.”
Total Focus
Silver met DeepMind CEO Demis Hassabis at a chess tournament when they were kids, and they became lifelong friends and collaborators. But Silver left DeepMind because he wanted a completely new direction. “I feel it’s really important that there is an elite AI lab that actually focuses a hundred percent on this approach,” he explains. “That it’s not just a corner of another place dedicated to LLMs.”
The limits of the LLM-based approach become clear, Silver says, with a simple thought experiment. Imagine releasing an LLM in a world that believed the Earth was flat. Without interacting with the real world, the system would remain an avid flat-earther, even if it improved its own code. In contrast, an AI that can learn about the world for itself could make its own scientific discoveries,and that, Silver believes, is the only path to genuine superintelligence.
(Source: Wired)
