Google DeepMind's boss on AI, power, God and what's next | The Economist

TL;DR

Demis Hassabis explains his scientific motivations for building AGI, dismisses 'building God' framing, and warns of non-zero catastrophic risk within ~5 years.

Key Points

• 1.Hassabis frames AI as a scientific tool, not a god. He rejects peers' 'building God' language, comparing AI to a telescope or microscope — an instrument to interrogate the fabric of reality, rooted in his lifelong obsession with understanding the universe.
• 2.Religion and science both shaped his worldview. Raised between a Baptist mother and atheist father, Hassabis sees science, philosophy, and religion as converging on the same questions about humanity's place in the universe, driving his 16+ years at DeepMind.
• 3.He acknowledges a non-zero chance AI could go badly wrong. Calling it 'cautious optimism,' he warns AGI could be as close as 5 years away and says the competitive race between companies and nations makes the international cooperation he originally envisioned — like a CERN for AI — nearly impossible today.
• 4.His life's mission remains getting AGI safely across the line. Despite winning a Nobel Prize — an item from his teenage plan — he says the most important work is still ahead, focused on ensuring AGI benefits all of humanity rather than causing harm.