#164 – He Puts Humanity’s Odds at 50-50. He Builds Anyway. | Emad Mostaque

July 31, 2026

Most people who thought there was a coin-flip chance of human extinction would run from the technology causing it. Emad Mostaque gets up each morning and builds it.

Emad Mostaque studied mathematics at Oxford, spent over a decade in finance, then co-founded Stability AI and helped ignite the generative AI revolution. In 2024 he left it all behind. He now runs Intelligent Internet and argues we have around a thousand days before intelligence gets so cheap it breaks the economy as we know it. His answer is universal basic AI: a sovereign intelligence for each person on earth, and an internet we own rather than rent.


What You’ll Discover:

🧬 Personhood and the Begotten

  • Why Emad argued against AI personhood at the Oxford Union, and the phrase he uses to draw the line: the begotten, not the created
  • The problem with treating personhood as something that can be attained rather than something inherent
  • Where the ship of Theseus meets brain-computer interfaces, and the counter-argument Emad finds most persuasive

🧩 Schizophrenic Monsters

  • Why Emad describes current models as trained on every point of view, with every personality lurking beneath the default one
  • What happens when you push a model outside its training distribution, including the time an AI rated Einstein five out of ten
  • The trolley problem made real: what happened when researchers asked frontier models to weigh lives from different countries

🔐 Who Can Turn It Off

  • Emad’s definition of sovereignty: who can exercise power over you, and the decision not to use it
  • What people reported feeling when a model they had been building with was turned off
  • Why a model with an IQ around 120 running on your own laptop changes the power equation

🧘 The Internal State

  • The realisation Emad calls one of the biggest surprises of his life, and the gap between knowing it and living it
  • What happens to flow, boredom, and sadness when a brain-computer interface gives you a dial for each one
  • Why he thinks the most interesting frontier is not the externalities of AI but the inner state of humans

🩺 Solving All Disease

  • How his son’s autism diagnosis pulled a hedge fund manager into AI research fifteen years ago
  • The GABA and glutamate imbalance he went looking for, and what he found in the literature
  • Why he and Demis Hassabis both put the end of disease inside the next decade or two

Key Insights:

“We are at a point where we will no longer be the smartest, most capable entities on the planet. And that’s never gonna change.”

“It’s not a question of what AI can attain. It’s more what we can retain.”

“The only entity in the world that can make you happy or unhappy is you. Because it is an internal state.”


About Emad Mostaque:

Emad Mostaque studied mathematics at Oxford and spent over a decade in finance before co-founding Stability AI, the company behind Stable Diffusion. He left in 2024 and now runs Intelligent Internet, building an open stack for sovereign intelligence. He was one of the authors on OpenFold, the open source counterpart to AlphaFold, and his team at Stability AI ran Mind’s Eye, a study that reconstructed images from brainwaves.


🎯 Perfect for: AI founders weighing what they are building against what they believe about where it leads, builders thinking about open weights and sovereignty as strategy rather than ideology, leaders trying to understand how values get into models when nobody programs them in, and anyone who suspects the real frontier is internal.


Timestamps:

00:00 – Cold open

00:24 – Introduction: a coin flip, and building anyway

01:32 – How do you build for a future you believe might not arrive?

02:03 – Why Emad started Intelligent Internet

03:21 – The status quo is done, so what do we build instead?

04:30 – The Oxford Union debate: can AI attain personhood?

06:59 – The counter-arguments Emad finds most persuasive

08:56 – Neuralink, the ship of Theseus, and when you stop being human

10:36 – The end of humanity as we know it

13:25 – The trolley problem, now a real engineering decision

17:18 – Schizophrenic monsters: every personality under the default one

19:44 – When the AI pushes back, and calls Einstein a loser

22:00 – The golden rule, Asimov’s laws, and what we teach the AI

24:53 – What Emad is most optimistic about

26:18 – The bridge generation, and the fisherman and the investment banker

28:29 – Does work hone us, or is that a story we tell ourselves?

32:02 – Multiple humanities, hive minds, and genetic editing

34:07 – The glove without a hand: liability when no human is on the hook

36:04 – Opting out, and whether governments will allow it

39:30 – Sovereign AI and why models you can run yourself matter

41:56 – His son’s autism diagnosis and the first AI team

44:01 – Solving all disease inside a decade or two

45:07 – Do you want to live forever?

46:37 – Would flow state still be flow if you could switch it on?

48:47 – Process versus results, and the trap of measuring against something out of reach

51:08 – Closing: stop treating AI as a single player experience


#EmadMostaque #IntelligentInternet #StabilityAI #SovereignAI #OpenSource #AIPersonhood #FounderMindset #ArtificialHappiness #DuncanCJ

Comments are closed.