🇬🇧Demis
Hassabis

Demis Hassabis learned chess at the age of four. By thirteen, he was the second-best junior chess player in the United Kingdom, rated master level, and already beginning to think about chess not merely as a game but as a model for structured reasoning under uncertainty — the kind of thinking he would eventually apply to protein structures, ancient games, and ultimately to the architecture of intelligence itself. At seventeen, having put chess aside to focus on something he found even more interesting, he co-designed Theme Park for Bullfrog Productions — a game that sold millions of copies and became a landmark in the simulation genre. He was still in secondary school.

After completing an undergraduate degree in computer science at Queens' College, Cambridge — which he finished in two years instead of three — Hassabis spent several years in the video game industry, designing and producing games of increasing complexity and ambition. But what he really wanted to understand was intelligence: not game intelligence or machine intelligence specifically, but intelligence as a general phenomenon. In 2004, he enrolled as a PhD student in neuroscience at University College London, studying how the hippocampus supports imagination and the simulation of future scenarios. The work won him the UCL Prize for outstanding doctoral thesis. He was developing, in parallel, a theoretical framework that would define his career: that the human brain was the best existing model of general intelligence, and that understanding it deeply enough would reveal principles that could be implemented in machines.

In 2010, Hassabis co-founded DeepMind with Shane Legg and Mustafa Suleyman in London — explicitly with the mission of building artificial general intelligence for the benefit of humanity. The company combined the neuroscience-inspired approach of Hassabis with machine learning engineering, and its early results were spectacular. In 2015, its AlphaGo system became the first AI to defeat a professional Go player, a feat that many experts had believed was a decade away. In 2017, AlphaGo Zero — which learned entirely through self-play, without any human game data — surpassed its predecessor within three days. Google acquired DeepMind for a reported $500 million in 2014, and Hassabis continued as CEO.

The culmination of DeepMind's scientific ambition came with AlphaFold. For fifty years, determining the three-dimensional structure of a protein from its amino acid sequence — the protein folding problem — had been one of biology's grand unsolved challenges. Protein structures determine function, and function determines medicine; knowing a protein's shape is essential for understanding disease and designing drugs. The critical assessment of protein structure prediction (CASP) competition had measured progress on the problem annually since 1994, with incremental improvements. In 2020, AlphaFold2 shattered the competition, predicting protein structures with accuracy comparable to experimental methods. In 2021, DeepMind made the AlphaFold database freely available to the world — 200 million protein structures, representing virtually every protein known to science.

The scientific impact was immediate and vast. Within months of release, researchers were using AlphaFold to accelerate work on antibiotic resistance, neglected tropical diseases, malaria, cancer, and dozens of other problems that had been blocked by the difficulty of determining protein structure. In 2024, the Nobel Committee awarded the Nobel Prize in Chemistry to Hassabis and John Jumper of DeepMind for AlphaFold, and to David Baker of the University of Washington for computational protein design. It was the first Nobel Prize explicitly awarded for work done primarily by an AI system — and for a British company that began as a startup in a London office a decade and a half earlier, led by a man who had learned chess at the age of four.