DeepMind

Demis Hassabis grew up as a child prodigy, becoming one of Europe's best chess players by age six and winning the under-eight championship. At a tournament in the mountains, a 30-year-old Danish champion humiliated him after a 10-hour match that should have been a draw. Faced with 300 brilliant minds competing in chess, young Demis had an epiphany: "If you took the 300 people in this room, the brainpower in this room, that we're just spending on this 10 hour tournament here, we could cure cancer." He abandoned chess and committed to harnessing human brainpower combined with computers to solve humanity's greatest problems. By age 16, while working at Bullfrog Productions on the game Theme Park, he pioneered AI logic for autonomous game characters—proving even then that AI was the only frontier worth pursuing.

Hassabis turned down a £1 million offer to stay at Bullfrog and instead attended Cambridge to study AI and neuroscience. There, he found a kindred spirit equally obsessed with teaching computers to think like human minds. Together they founded DeepMind when AI was dismissed by both science (no testable hypotheses) and entrepreneurship (pure sci-fi). Peter Thiel became the first believer, followed by Elon Musk. They started with a radical premise: don't tell the AI the rules. Just give it a reward signal. They trained their system on Atari games—Pong, Brick Breaker—with only one instruction: "a higher score is good." Within 500 games, the AI surpassed human performance and invented strategies humans never discovered.

Google acquired DeepMind for approximately £400 million (roughly $500+ million USD), recognizing the company's AI as their strategic weapon. This validation launched the era of AI-assisted science. Hassabis turned attention to protein folding—a problem unsolved for 50 years. The CASP competition measured progress at 20-30% accuracy for over a decade. DeepMind won the first year but fell short of the 90% accuracy needed to truly solve it. Rather than panic, Hassabis gave his team creative space: "When you need creative ideas, you can't force them. Squeezing doesn't make creativity come out." He identified the J-curve moment—when new approaches initially decline before breaking through—and pushed at exactly the right time.

The breakthrough came from AlphaZero's approach: train on zero human knowledge, only self-play. Applied to protein folding, the accuracy graph exploded from 30% to 90% in one year. Move 37 in AlphaGo vs. Lee Sedol became legendary—an entirely novel move no human would make, signaling the moment AI stopped mimicking humans and started creating. The Chinese government cut the broadcast when AlphaGo beat their number-one player, triggering China's AI arms race. These weren't incremental wins; they were inflection points. Hassabis's conviction never wavered: "Today the whole world agrees with something that I knew 20 plus years ago, that AI is the most important technology that we're ever going to build."

DeepMind spawned Isomorphic Labs, a Google-backed company with a $600 million founding round, tasked with curing all disease through AI-discovered drugs. The protein folding breakthrough unlocked drug discovery at scale—simulating efficacy, predicting protein structure, targeting disease at the molecular level. Hassabis remains obsessed: "How will we get sleep? I won't be able to sleep." When offered billions to exit, he calculated the trade-off—a few billion dollars for five more years to see AGI in his lifetime. He chose the mission. The documentary captures him not at the finish line, but at the beginning of something far larger: using AI to solve the problems that have haunted humanity for millennia.