Intelligence Briefing
Led the development of AlphaGo, AlphaZero, and AlphaStar at Google DeepMind, achieving landmark results in Go, chess, shogi, and StarCraft II. One of DeepMind's first employees (2010). Left DeepMind in late 2025 after a sabbatical to found Ineffable Intelligence in London, aiming to build "an endlessly learning superintelligence that self-discovers the foundations of all knowledge." Reportedly raising $1B in a seed round led by Sequoia Capital, which would be Europe's largest-ever seed round.
BA, Computer Science — University of Cambridge
MA, Computer Science — University of Cambridge
PhD, Reinforcement Learning — University of Alberta
Operational History
Left Google DeepMind
Departed from Google DeepMind after a sabbatical to focus on new ventures.
departureFounded Ineffable Intelligence
Established Ineffable Intelligence in London, focusing on developing superintelligence.
foundingAlphaStar Achieves Grandmaster Level
AlphaStar, developed by his team, achieves Grandmaster level in StarCraft II.
researchAlphaZero Introduced
Introduced AlphaZero, a general reinforcement learning algorithm that mastered chess, shogi, and Go.
researchAlphaGo Defeats Lee Sedol
AlphaGo, developed under his leadership, defeats world champion Go player Lee Sedol.
researchJoined Google DeepMind
Became one of the first employees at Google DeepMind.
careerAGI Position Assessment
2030-2040
Advocates for a cautious approach to AGI development, emphasizing safety and self-discovery.
- Self-play is essential for safe AGI.
- Human feedback alone is insufficient for AGI development.
Focus on self-discovery and iterative learning.
Intercepted Communications
“We are building an endlessly learning superintelligence that self-discovers the foundations of all knowledge.”
“Self-play is crucial for developing safe AI systems.”
“The future of AI will not be defined by human feedback alone.”
“AlphaGo was just the beginning; we have much more to achieve.”
“I believe in the potential of AI to revolutionize our understanding of intelligence itself.”
Research Output
The Importance of Self-Play in Reinforcement Learning
2021arXiv
Discussed the critical role of self-play in training AI systems.
A Survey of Deep Reinforcement Learning
2020arXiv
Comprehensive overview of advancements in deep reinforcement learning.
A General Reinforcement Learning Algorithm that Masters Chess, Shogi, and Go through Self-Play
2019Proceedings of the National Academy of Sciences
Further established the effectiveness of self-play in AI.
AlphaStar: Mastering Real-Time Strategy Games through Deep Reinforcement Learning
2019Nature
Showcased the application of reinforcement learning in complex real-time strategy games.
Mastering Chess and Shogi by Self-Play with a General Reinforcement Learning Algorithm
2018Science
Demonstrated the power of self-play in reinforcement learning.
Mastering the Game of Go with Deep Neural Networks and Tree Search
2016Nature
Pioneering work in reinforcement learning and game AI.
Deep Reinforcement Learning with Double Q-learning
2015AAAI
Introduced Double Q-learning, a significant advancement in reinforcement learning.
Playing Atari with Deep Reinforcement Learning
2013arXiv
Pioneered deep reinforcement learning techniques in gaming.
Known Associates
Demis Hassabis
colleagueCo-founder and CEO of Google DeepMind, worked closely with David on various projects.
View Dossier →John Schneider
collaboratorCollaborated on the development of AlphaGo and other AI projects.
View Dossier →Olivier Vinyals
colleagueWorked with David on AlphaStar and other reinforcement learning initiatives.
View Dossier →Richard Sutton
mentorInfluential figure in reinforcement learning, provided guidance during David's early career.
View Dossier →Organizational Affiliations
Current
Ineffable Intelligence
CEO & Founder
2025-present
Former
Google DeepMind
Principal Research Scientist
2010-2025
University College London
Professor
2015-2020
Source Material
Dossier last updated: 2026-03-04