all figures

Demis Hassabis

戴密斯·哈萨比斯

AI research · Reinforcement learning · Protein structure

Kernel

Hassabis is the figure who, more than any other, has carried the British research tradition into the AI era. Child chess prodigy, video-game designer (Theme Park, 1994), neuroscience PhD at UCL, founded DeepMind in 2010, sold it to Google in 2014. AlphaGo's 2016 defeat of Lee Sedol; AlphaFold's 2020–24 solution of protein structure prediction at near-experimental accuracy; the 2024 Nobel Prize in Chemistry for the latter.

§ 01

Contribution

AlphaGo and its successors (AlphaZero, MuZero) as the canonical proof that reinforcement learning can master complex domains. AlphaFold's protein-structure prediction (now used by every structural biology lab on earth). Gemini's frontier-model family. A research culture that preserves academic seriousness inside a Google subsidiary.

§ 02

Lineage

UCL neuroscience (Eleanor Maguire). The British AI tradition that runs from Turing → I. J. Good → DeepMind's early hires. Larry Page's 2014 acquisition placed DeepMind inside Google's compute and capital substrate, which permitted the AlphaFold scale. The lineage now seeds Isomorphic Labs and a generation of AI-for-science startups.

§ 03

Civilization-scale significance

The 2024 Nobel for AlphaFold is the first Nobel won by a working machine-learning system. It is also the formal endorsement by the Swedish Academy that frontier AI has crossed into the natural-science contribution category.