Demis Hassabis (哈萨比斯): the scientist-CEO chasing AGI while winning science’s top plaudits

From prodigy to scientist-CEO

Demis Hassabis (哈萨比斯) has been framed as a rare blend of rigorous scientist and ruthless executor — a chess prodigy turned game designer, neuroscientist and the founder who built DeepMind into one of the world’s most consequential AI labs. It has been reported that Hassabis won the 2024 Nobel Prize in Chemistry for breakthroughs in protein-structure prediction, a capstone to a career that also produced AlphaGo and AlphaFold. Can one person be both a Nobel laureate and the CEO driving an effort to build artificial general intelligence (AGI)? Hassabis’s trajectory suggests yes, and that raises both excitement and hard questions.

Milestones and method

Hassabis’s CV is familiar to AI watchers: child chess master, Cambridge computer science undergraduate, a UCL neuroscience doctorate studying the hippocampus, and co‑founder of DeepMind in 2010. Under his leadership DeepMind developed AlphaGo — the system that defeated top human go players in 2016–17 — and AlphaFold, which transformed protein-folding prediction and reportedly opened new paths in biology and medicine. He has been described by peers as a polymath who moves quickly between disciplines, and it has been reported that the late Stephen Hawking once called him “one of the smartest people” he had met.

Geopolitics, compute and the China question

Hassabis’s work sits squarely in the middle of a geopolitical contest over AI. Google’s 2014 acquisition of DeepMind and the 2023 reorganization that merged Google Brain with DeepMind reflect tech consolidation in an era of rising export controls and chip restrictions. Hassabis has reportedly told interlocutors that China’s frontier teams — including models like DeepSeek and firms such as Alibaba (阿里巴巴) — may be only months behind leading U.S. groups, even as Beijing still lags on leading-edge compute because of Western export controls. That compute gap, analysts say, is a strategic choke point in the race to AGI.

Vision, safety and the road ahead

Hassabis frames AGI as a scientific project as much as an engineering one: he has reportedly put a 50% probability on AGI by 2030, but insists true AGI must be able to propose scientific hypotheses and understand the physical world, not just excel at benchmark tasks. He has also advocated for international cooperation — a CERN‑style body for AI research — arguing that AGI safety is a global public‑goods problem that no single company or country can solve alone. Whether the coming decade delivers the breakthroughs he predicts, Hassabis’s mix of deep science and industrial scale will remain central to how the world answers the biggest questions about intelligence and risk.