David Silver Raises $1.1B for AI Lab Without Human Data

Ineffable Intelligence, a new AI company in Britain started just months ago by David Silver from DeepMind, collected $1.1 billion in funding. The investment came at a $5.1 billion valuation. The firm enters the competition to develop new AI models that might surpass large language models.

The company's website states its goal clearly. Ineffable wants to produce a superlearner. This system would uncover knowledge and abilities on its own. It would use reinforcement learning, where AI improves by trying actions and learning from results. No human-created examples would be needed. Reinforcement learning matches Silver's main field of work.

Silver's Career at DeepMind and Beyond

David Silver teaches as a professor at University College London. Until lately, he headed the reinforcement learning group at DeepMind, owned by Google. He worked there for over ten years before starting this company.

During his time at DeepMind, Silver helped create AI that defeated expert humans in chess and Go. These programs learned only from playing games themselves. They did not use human tactics or past game data. This approach beat the best existing computer systems in both games. AlphaZero stands out as the prime example. Ineffable plans for its superlearner to find all knowledge through personal experience, much like that.

DeepMind, founded in London in 2010 by Demis Hassabis and others, gained fame for such advances. Google bought it in 2014. Silver led key projects like AlphaGo, which in 2016 beat world Go champion Lee Sedol. AlphaZero built on that in 2017, mastering chess, shogi, and Go from scratch in hours.

Bold Claims and Personal Commitments

The superlearner starts with no prior experience. Yet Ineffable shows great drive. Its site declares that success would match the scale of Charles Darwin's discovery. Darwin's principle covered all life. Ineffable's principle would cover and create all intelligence. The statement uses capitals for emphasis.

Silver described Ineffable as his life's work in a note on the company blog. He also spoke to Wired. He said any profits from the company would support charities. Those charities focus on saving the most lives possible.

Details on revenue remain unknown. No one knows how, when, or how much money the company might earn. That uncertainty did not stop investors from committing funds.

Investors and Valuation Milestone

Wired reported the funding round's leaders. Sequoia Capital and Lightspeed Venture Partners took the lead. Index Ventures, Google, Nvidia, and more joined. Other backers include the British Business Bank and Sovereign AI. Sovereign AI is the U.K.'s new fund for national AI efforts.

Ineffable reached pentacorn level right away. That term applies to firms valued over $5 billion. It joins other AI startups by top researchers. Their fame draws huge early funds, called coconut rounds in jest, beyond standard seeds.

Last month, AMI Labs got $1.03 billion at a $3.5 billion pre-money valuation. Yann LeCun co-founded it. LeCun won the Turing Award and worked at Meta AI.

Recursive Superintelligence also fits this pattern. Tim Rocktäschel, once DeepMind's principal scientist, co-founded it in the U.K. It raised $500 million. Demand could push that to $1 billion.

Growing AI Presence in London

These firms, with U.S. links too, point to London's rise as an AI center. DeepMind's base there persists post-Google buyout. Jeff Bezos's Project Prometheus seeks space near Google's AI site.

A strong network of former DeepMind employees exists. Several will join Ineffable's leadership team.

Related on Neura Market

• AI Tools Directory
• AI Startups
• Reinforcement Learning Resources