DeepMind’s David Silver raises $1.1B for AI that learns without human data
▼ Summary
– Ineffable Intelligence, a British AI lab founded by former DeepMind researcher David Silver, raised $1.1 billion at a $5.1 billion valuation to develop AI models beyond large language models.
– The lab aims to create a “superlearner” using reinforcement learning, where AI learns through trial and error without relying on human data.
– Silver previously led DeepMind’s reinforcement learning team and contributed to programs like AlphaZero that mastered chess and Go through pure self-play.
– The funding round was led by Sequoia Capital and Lightspeed Venture Partners, with investors including Google, Nvidia, and the U.K.’s Sovereign AI fund.
– Ineffable is part of a trend of large “coconut rounds” for AI ventures founded by star researchers, highlighting growing momentum for London as an AI hub due to DeepMind’s influence.
A new British AI lab, Ineffable Intelligence, has secured $1.1 billion in funding just months after its founding, reaching a valuation of $5.1 billion. The company was established by former DeepMind researcher David Silver, who is betting on a radical departure from today’s dominant large language models.
The startup’s mission, as outlined on its recently launched website, is to build a “superlearner.” This system would acquire knowledge and skills through reinforcement learning, a methodology where AI improves via trial and error rather than by ingesting human-generated data. Silver, a professor at University College London, spent over a decade at Google-owned DeepMind leading its reinforcement learning team before launching this venture.
At DeepMind, Silver helped create programs like AlphaZero, which mastered chess and the board game Go purely through self-play, without any human strategies or game records. Ineffable Intelligence aims to replicate that success on a far grander scale, envisioning an AI that discovers all knowledge from its own experience.
The company’s ambitions are audacious. Its website claims, “If successful, this will represent a scientific breakthrough of comparable magnitude to Darwin: where his law explained all Life, our law will explain and build all Intelligence.” Silver has described Ineffable as “his life’s work” in a personal blog post. He told Wired that any profits he earns will go to high-impact charities. How or when the venture will generate revenue remains unclear, but that uncertainty has not deterred investors.
The funding round was led by Sequoia Capital and Lightspeed Venture Partners, with participation from Index Ventures, Google, Nvidia, the British Business Bank, and Sovereign AI, the U. K.’s new sovereign venture fund for AI. The massive early-stage investment places Ineffable among a growing group of ventures founded by star researchers that have attracted such large seed rounds they are informally called “coconut rounds.” Last month, AMI Labs, co-founded by Turing Award winner Yann LeCun, raised $1.03 billion at a $3.5 billion pre-money valuation.
Another example is Recursive Superintelligence, co-founded by former DeepMind principal scientist Tim Rocktäschel, which reportedly secured $500 million with demand sufficient to push that figure to $1 billion. These developments underscore London’s rising prominence as an AI hub, a trend driven by DeepMind’s continued presence following its 2014 acquisition by Google. Jeff Bezos’ AI lab, Project Prometheus, is also reportedly scouting office space near Google’s AI hub.
Ineffable’s executive team is expected to include several former DeepMind staffers, leveraging a powerful alumni network. For now, the company has a bold vision, a massive war chest, and a clear bet that the next leap in intelligence will come not from human data, but from machines teaching themselves.
(Source: TechCrunch)
