Ineffable Intelligence

A London AI research start-up formed in late 2025 and led by David Silver, a leading reinforcement-learning researcher who helped build AlphaGo. It aims to build AI that keeps learning by itself.

About Ineffable Intelligence

Ineffable Intelligence is a new London-based AI start-up working on one of the most ambitious goals in the field: building what its founder calls "an endlessly learning superintelligence that self-discovers the foundations of all knowledge". It is recruiting AI researchers, and because it is so new, there is not yet much public detail about its products.

It was formed in November 2025 by David Silver, who left Google DeepMind at the end of that year; Fortune reported the move in January 2026. Silver joined DeepMind shortly after it was founded in 2010 and contributed to AlphaGo, AlphaZero, MuZero and AlphaStar, and he played a central role in building Google's Gemini models. He is also a professor at University College London.

The company's approach is different from the large language models behind today's chatbots, which learn from human-written text. Ineffable Intelligence plans to rely on reinforcement learning, where AI systems learn through trial and error, often in simulated environments. Silver argues that learning only from human knowledge limits AI, and the approach centres on "world models" that let an AI predict the results of its actions and keep adapting.

Our sources do not name other team members, publish a company mission page or describe offices beyond London, so we have kept to what has been reported. We have left out funding.

For young people, Ineffable Intelligence is a window into frontier AI research. The skills behind this kind of work are mathematics, computer science and machine learning, and much of the field comes through university research. If this interests you, look at university AI and machine learning courses and follow the company's announcements for future opportunities.

What they do

• {'desc': 'AI that learns by trial and error.', 'title': 'Reinforcement learning research'}
• {'desc': 'AI that predicts the results of its own actions.', 'title': 'World models'}
• {'desc': 'Long-term research towards continually learning AI.', 'title': 'Frontier AI research'}

Key facts

• Based in: London, UK
• Formed: November 2025
• Founder: David Silver, professor at UCL and former DeepMind researcher
• Stated goal: an endlessly learning superintelligence
• Approach: reinforcement learning rather than learning from human text