AI leaders like Dario Amodei and Elon Musk are predicting that AI will drive GDP growth into double digits—somewhere between 10% and 100% annually in the 2030s. These aren't fringe predictions. They're common among AI insiders, and Anthropic's own economic model reaches double-digit growth in its "extreme" scenario. But the authors, who are bullish on AI capabilities, argue this won't happen in the next 10-15 years. They've even bet money on it: US per capita real GDP growth will stay below 15% annually through 2033. Their reasoning starts with a simple reframing. Instead of thinking about growth rates directly, ask yourself how much richer you'll be in 15 years. If you think we'll be twice as rich, that's only 4.7% annual growth—already massive. A 16.6% growth rate would mean we're ten times richer. These aren't modest predictions; they're claims about a hundredfold increase within 30 years. The authors acknowledge that explosive AI-driven growth is theoretically possible. Standard growth models, when you plug in reasonable assumptions about automation replacing labor and eliminating bottlenecks, can produce double-digit growth. There's nothing mathematically wrong with the idea. But theory and reality diverge. The predictions rest on five assumptions that likely won't all hold simultaneously. Fast, economy-wide automation needs to happen everywhere—not just in tech. Consumers need to keep demanding and buying whatever gets automated. Someone has to fund all that production. AI-driven cyberattacks can't destroy value faster than AI creates it. And AI has to accelerate R&D itself to sustain explosive growth. Each assumption faces real-world friction. Automation doesn't spread evenly. People value services that can't be automated—care work, relationships, scarce physical goods. Investment doesn't automatically follow capability improvements. The authors think people extrapolate from their immediate environments and miss what's happening in the broader economy. They see AI breakthroughs in their field and imagine the whole economy transforming at the same pace. There's also motivated reasoning—people want AI to be transformative and think backward from that desire. The elegant mathematics of growth theory seduces people into believing models that assume away the messy constraints of actual economies. A capabilities explosion and a GDP explosion are different things. One is about what AI can do. The other is about whether those capabilities translate into sustained, economy-wide productivity gains that nobody's going to block or redirect toward things that don't show up in GDP measurements.