After leading the AlphaGo and reinforcement-learning research at DeepMind for over a decade, he left in 2025 to found Ineffable Intelligence and build a superlearner that discovers knowledge from its own experience without human data.