Sebastian Mallaby

2 episodes · 10 moments · last on 8 Apr 2026

8 Apr 2026FULL INTERVIEW: Sebastian Mallaby on Sam Altman, OpenAI and Silicon Valley

0:002 min

Who Is Demis Hassabis? Mallaby's Portrait of a 'Good Guy' in AI

The hosts introduce guest Sebastian Mallaby, author of The Power Law and More Money Than God, to discuss his new book on Demis Hassabis. Mallaby says he spent over 30 hours with Demis and spoke to roughly 100 people around him, concluding that Demis is a good guy: messianic about bringing AI to humanity, but with strong values. He notes Demis stayed in Britain, citing American inequality and 'fake it till you make it' culture, and is a loyal family man married to his college sweetheart with two kids.

2:072 min

Gödel, Escher, Bach and the Brain-as-Computer Belief

Asked why Gödel, Escher, Bach influenced Demis so deeply, Mallaby says its central idea — that the brain is like a computer, with synapses firing ones and zeros — gave Demis the conviction that computers could eventually do anything brains can. Demis embraced this in the mid-1990s, 15 years before AI achieved much of anything. Mallaby notes Demis founded DeepMind in 2010, five years before Sam Altman and Elon Musk started OpenAI in 2015.

3:394 min

OpenAI's Founding Narrative and Musk's Double-Cross of DeepMind

The hosts note the irony that OpenAI branded itself as open and safe while DeepMind was painted as big-tech evil. Mallaby recounts that Demis sold DeepMind to Google in 2014 only after securing a safety and ethics oversight board, having turned down more money from Facebook and refused to sell to Elon Musk. Musk then hosted the board's first secret meeting, learned DeepMind's plans, and launched OpenAI with Sam Altman as a competitor — which Mallaby attributes to the big tech egos jealous of Demis's leadership.

7:245 min

Is AI in Safe Hands? Race Dynamics, the Pentagon, Washington and China

Mallaby argues the core AI safety problem is bad dynamics rather than bad individuals: Demis is a good person trapped in a vicious AGI race where unilateral restraint accomplishes nothing. He cites Anthropic refusing Pentagon uses like mass surveillance and autonomous lethal weapons, only for Sam Altman to undercut Dario Amodei by offering the Pentagon whatever it wants. He says AlphaFold and Demis's Nobel Prize show that life-saving drugs could shift opinion — two-thirds of Americans currently see AI as threatening — but that safety is impossible while the Trump administration ignores AI governance. After eight days in China, he found Chinese labs and CEOs openly discussing safety, making a non-proliferation-style treaty with controls on open source conceivable.

11:582 min

Musk vs. Larry Page and the 'Speciesist' Comment

On the book's viral Musk–Larry Page exchange, Mallaby says Page genuinely follows the logic that what matters in evolution is the survival of intelligence — even silicon intelligence over biological — and doesn't seem troubled that flesh-and-blood intelligence might be surpassed or eliminated. That attitude drove Musk crazy and prompted their falling out. Mallaby adds that the earliest AI pioneers are all unusual people; Demis — a self-taught prodigy who entered Cambridge at 16 — is unusual but fortunately good, while others are not.

14:122 min

DeepMind's Secret Hedge Fund

The host raises the book's brief mention of a secret hedge fund at DeepMind. Mallaby says Demis never told him directly — others around him did, and Demis acknowledged it only reluctantly. He recalls meeting Demis at European tech conferences and intriguing him with his Renaissance Technologies connections, including Geoffrey Hinton's first PhD student who built an early quant fund; DeepMind insiders later confirmed the effort to build a trading operation.

16:294 min

DeepMind and Google: From Rocky Years to the Pichai–Hassabis Era

Mallaby defends the 2014 sale as a British triumph that pulled nearly a billion dollars a year of American capital into UK R&D. The relationship soured after the oversight board collapsed, and Demis and Mustafa Suleyman fought a three-year secret battle (Project Mario) for an independent safety board, backed by a $1 billion spin-out offer from Reid Hoffman — ending with Suleyman pushed out around 2018–19. After ChatGPT launched in November 2022, Sundar Pichai merged Google Brain with DeepMind under Demis, starting a golden age; Mallaby calls the Pichai–Hassabis double act the most important partnership in global business.

20:162 min

Why 2 A.M. Is Demis's Favorite Time — and Closing

In the closing question, Mallaby explains that 2 a.m. is when Demis escapes the noise of the Silicon Valley AI race and his CEO duties to think and do science, reflecting his split persona between competitive capitalism and Nobel-level research. Demis told him that doing science gives him a spiritual sensation — that nature's intelligibility suggests some intelligent divinity may be behind it — his version of God. The hosts thank Mallaby for his time and wrap the conversation.

8 Apr 2026TBPN Acquired, New Yorker Vs. Sam Altman, Author of The Infinity Machine, Citrini Field Trip

1:45:5515 min

Sebastian Mallaby on Demis Hassabis & 'The Infinity Machine'

Author Sebastian Mallaby discusses his book on Demis Hassabis, calling him a good guy with strong values after 30+ hours of interviews and conversations with 100 people around him. He covers Hassabis's Gödel, Escher, Bach-inspired conviction that brains are computers, DeepMind's safety-and-ethics board condition on the Google sale, Musk attending the first oversight meeting then launching OpenAI as a rival, race dynamics that make lab-level safety impossible (citing Anthropic vs. the Pentagon), safety conversations in China, and Larry Page's "speciesist" remark that fueled the Musk falling-out. The segment continues in part 2.

2:00:378 min

Sebastian Mallaby: DeepMind's Secret Hedge Fund & the Google Marriage

The interview with Demis Hassabis biographer Sebastian Mallaby continues. He explains how DeepMind secretly built a trading operation (which Hassabis never told him about directly; he learned via contacts from his Renaissance Technologies book), then walks through the failed safety-and-ethics commission with Elon Musk, Hassabis and Mustafa Suleyman's three-year "Project Mario" fight to force an independent safety board on Google — including a $1 billion Reed Hoffman spin-out offer — and the post-ChatGPT merger of Google Brain into DeepMind that made the Pichai–Hassabis duo "the most important buddy act in global business." It closes with why 2 AM is Hassabis's favorite hour for science and quasi-spiritual thinking, before the hosts thank Mallaby and plug The Infinity Machine.