The future of intelligence | Demis Hassabis (Co-founder and CEO of DeepMind)

In this episode of the Google DeepMind podcast, Professor Hannah Fry interviews Demis Hassabis, CEO and co-founder of Google DeepMind, about the current state and future directions of AI. They discuss recent advancements like Gemini 3 and world models, as well as the potential of AI to solve "root node" problems such as fusion and material science. Hassabis addresses the paradox of AI excelling at complex tasks while making basic errors, the need for consistency and reasoning in AI systems, and the balance between commercial applications and scientific research. They explore the importance of simulated worlds for understanding physics and potentially consciousness, the societal impact of AGI, and the need for international collaboration to manage its development responsibly. Hassabis also shares personal reflections on the challenges and excitement of leading AI research, and his vision for a future where AI helps humanity steward AGI safely.