British AI researcher David Silver is pursuing one of the most ambitious goals in artificial intelligence: building an “endlessly learning superintelligence that self-discovers the foundations of all knowledge.” His new London-based startup, Ineffable Intelligence, has reportedly raised $1 billion in a seed round at a valuation of approximately $4 billion — potentially the largest seed round ever for a European startup.
Silver left Google DeepMind last year, a move that reportedly triggered intense competition among investors. According to the Financial Times, the round was led by Sequoia Capital, with additional interest from Nvidia, Google, and Microsoft.
From AlphaGo to a New AI Paradigm
At DeepMind, Silver was a central figure behind AlphaGo, the reinforcement learning system that defeated world champion Lee Sedol in 2016. He also contributed to AlphaZero and played a role in developing Google’s Gemini family of models.
But Ineffable Intelligence represents a philosophical shift. Silver argues that today’s large language models are fundamentally constrained because they are trained primarily on human-generated data. No matter how large the dataset, such systems remain bounded by existing human knowledge.
His alternative is reinforcement learning from experience. Instead of absorbing static text corpora, AI systems would learn through trial and error, interacting with environments and refining internal “world models” that predict the consequences of their actions.
The “Era of Experience”
In a paper co-authored with computer scientist Richard Sutton, Silver described the coming “Era of Experience,” in which AI agents achieve superhuman capabilities by learning directly from interaction rather than imitation. Over time, experience would become the dominant driver of improvement — potentially surpassing the value of all human-generated data combined.
The vision is not a model trained once and deployed, but a continuously adapting system that evolves for months or years. Such agents could autonomously tackle complex tasks, discover new strategies, and generate knowledge beyond the limits of human expertise.
By raising $1 billion at inception, Ineffable Intelligence signals that investors believe this reinforcement learning-driven approach could define the next phase of AI development. If successful, Silver’s bet would mark a transition from AI systems that reflect what humanity already knows to systems capable of discovering what humanity does not.
For Silver, the goal is clear: not just smarter models, but a fundamentally new kind of intelligence.
