Freakonomics Radio

682. Should A.I. Move to Space?

48 min

The episode explores the idea of moving AI computing to space to address the massive energy demands of data centers. Blaise Aguera y Arcas and his team at Google's Paradigms of Intelligence project argue that space-based solar power, combined with lightweight satellite data centers, could be more efficient and sustainable than terrestrial alternatives. The conversation also delves into the nature of intelligence, the social aspects of AI, and the practical challenges of building orbital data centers.

Summarized by Podsumo

🎧 Listen 🎙️ Ask about this episode

✨ Key Takeaways

1

Blaise Aguera y Arcas discusses his 'Paradigms of Intelligence' team at Google, which studies the social nature of intelligence, both in humans and AI, and how inner dialogue and disagreements can improve problem-solving.
2

The episode presents 'Project Suncatcher', an ambitious plan to put AI data centers in space on lightweight 'dragonfly-like' satellites, powered by constant sunlight in sun-synchronous orbits.
3

A key insight is that space-based solar power can gather up to eight times the energy of ground-based panels, and that moving computation to space avoids the inefficiencies of beaming energy back to Earth.
4

The economics hinge on launch costs falling to approximately $200 per kilogram, a milestone that seems increasingly plausible given the trend of declining launch costs, largely driven by SpaceX.
5

Will Marshall, founder of Planet Labs, argues that space-based data centers are inevitable within a decade, citing both economic and sustainability benefits for preserving Earth's biodiversity.
6

Blaise Aguera y Arcas proposes that intelligence is fundamentally about prediction, and that consciousness may emerge from the brain's ability to model itself and its own actions.
7

The discussion highlights the surprising insight that AI models, like human brains, exhibit an internal 'society of subagents' that argue with each other, similar to the Gollum/Smeagle dynamic in 'Lord of the Rings'.
8

Planet Labs, the company prototyping the satellites for Project Suncatcher, currently monitors the entire Earth's landmass daily, providing data for agriculture, disaster management, and security, while navigating ethical dilemmas around surveillance.
9

The episode reveals that Google's CEO Sundar Pichai and co-founder Sergey Brin were supportive of the seemingly 'crazy' idea, with Sergey Brin even attending the initial pitch meeting.
10

The need for a massive scale of satellites is emphasized, with the area of solar panels required being a 'tiny, tiny drop in the bucket' compared to the vastness of low Earth orbit.

💬 Notable Quotes

"The moment you start to have something that behaves intelligent, the question like, 'well, but is it really intelligent' is an interesting one to ask... I'm kind of with Turing on that one. If all of the tests check out, then that is your answer."

— Blaise Aguera y Arcas
"What people might find confusing about this, though, is that we've built these neural nets... and it is bizarre that the objective is to predict the next word. And I think what you're arguing is that you can't predict the next word well at all unless you have a pretty complete model of how the entire world works."

— Steve Levitt
"The insight behind something like Suncatcher is like, hey, space is a great place to do solar power, but with AI, maybe we actually have a really important and useful way to use that power in space. The thing that's different about AI versus some other applications... you just need to send data up and down to these satellites."

— Travis Beals
"I actually think it's quite clearly going to happen... We did a calculation with Google about eight or nine years ago now... and figured out that by around cost of launch coming to $200 to $300 a kilogram, it would just be cheaper. Purely on cost grounds, it would be cheaper to put them in orbit than to put them on the ground."

— Will Marshall