My First Million · Episode Brief
Story Of The Most Important Founder You've Never Heard Of
Demis Hassabis built DeepMind to solve intelligence, and the story of how he got there is more interesting than any of the AI hype that followed.
Demis Hassabis is the kind of person who makes the rest of the tech industry feel like they're playing a different game. Child chess prodigy, game designer, neuroscience PhD, AI researcher, and now one of the most consequential scientists alive. Sam and Shaan tell his story with the affection they reserve for founders who had no business succeeding and succeeded anyway.
The opening 'Demis the Menace' segment covers the early biography quickly but doesn't rush past what made him unusual: the combination of deep technical capacity with a genuine theory of what he wanted to use it for. Most technical prodigies optimize for the next impressive thing. Hassabis from an early age was trying to solve a specific problem — understanding intelligence — and everything else was instrumental to that.
Move 37 is the pivot point of the episode. The AlphaGo move that no human player would have made, that professionals called a mistake before realizing it was decisive — Shaan uses it as evidence for the broader claim that AI systems operating outside human heuristics have already produced genuine novelty. Not simulation. Not optimization on human training data. Something new.
The AlphaFold protein folding breakthrough is treated as the clearest proof of concept that Sam and Shaan have ever discussed on the show. A 50-year problem in biology — solved. Not improved upon, solved. The 'we are the gorillas' frame at the end is the most honest reckoning the show has done with what it means to be a non-technical observer of what's being built.
Key Ideas
- →Hassabis built DeepMind around a specific thesis — that solving intelligence algorithmically would unlock progress across every domain of science — rather than around a product or a market.
- →Move 37 in the AlphaGo vs. Lee Sedol match is used as evidence that AI systems have already produced genuinely novel outputs that were invisible to human expertise until after the fact.
- →AlphaFold's protein folding solution is framed as a category-change: not incremental improvement but the actual solution to a problem that had stumped biologists for fifty years.
- →Resourcefulness as the only resource that scales: Shaan's argument that the founders who navigate genuine resource constraints develop judgment that well-funded founders systematically miss.
- →The 'we are the gorillas' framing: humans may already be in the position of the animal that is intelligent enough to sense its own relative intelligence is not the highest in the room.
Worth Remembering
The Move 37 story told in real time — the professionals watching the game assuming the AI had made an error, then the slow realization that they were watching something they couldn't understand.
Sam's audible discomfort with the 'we are the gorillas' concept — genuine and unresolved rather than performed.
The childhood biography: Hassabis as a teenager simultaneously ranked among the top chess players in the world and designing video games that sold millions of copies.
The Paul Graham 'fierce nerds' essay referenced as the best one-page description of the people who build things that actually change the world.