Why AI’s Mathematical Breakthroughs Won't Trigger a Job Apocalypse
By Bram de Vries · Reporting from Amsterdam ·
Intellectual milestones in mathematics and AGI are high-status wins that fail to disrupt the practical realities of physical labor.
The latest "TBPN" podcast reveals a surprising truth: the "AI jobs apocalypse" has stalled. The Economist reported that AI created roughly one million American jobs—mostly in server farm construction—while causing 200,000 layoffs since mid-2023. In a climate of panic, the only figure that matters is that AI-driven job losses account for only one in every hundred shifts in the US workforce.
Why Math Prizes Don't Matter
The speakers spent much of the episode debating AI’s recent wins in mathematics. The systems earned gold medals at the 2025 International Math Olympiad and began chipping away at the Millennium Prize problems. Tyler argued that solving the Navier-Stokes equations is the ultimate intellectual achievement, a masterpiece of pure thought. The host disagreed, calling these breakthroughs marketing stunts. He noted that engineers already solve these equations numerically to design airplane wings and forecast weather.
The host is right. In my years in shipping and trade finance, I learned that a proof on a chalkboard is a luxury, but a functioning system is a necessity. Chasing Millennium Prizes is a high-status game for researchers at Google and Anthropic. These wins are intellectually stimulating, but they do not lower the cost of a container shipment or speed up a port. The host predicts the hype will simply migrate to biotech and cancer cures once the math prizes are won.
Can a Robot Paint a Bridge?
The conversation then shifted to Artificial General Intelligence (AGI). Josh (Sentry) said that after the Astra system scored 99% on ARC AGI 3, AGI is no longer a fantasy. This is the core of the apocalypse argument: once a machine reasons across any domain, it erases the human record of labor.
This logic fails when it hits the pavement. The episode described Astra controlling a robot to paint the Golden Gate Bridge and generating Lego kits from text prompts. These are incremental gains in robotics and 3D design, not a replacement for human utility. The job data tells the real story. Growth is not in "infrastructure," but in pouring concrete for data centers and laying miles of fiber-optic cable. We are not seeing the end of work, but the growth of the industrial base needed to power the compute.
The Industrial Hunger
Technology shifts labor; it does not delete it. From the steam engine to the mainframe, the pattern holds. Alarmists view the economy as a fixed pie, while the actual growth comes from those building bigger kitchens. To call this shift "unprecedented" ignores two centuries of factories and assembly lines.
The obsession with AGI and math proofs distracts the Silicon Valley elite. The real story is that the labor market is holding firm because the machines require a massive physical footprint. We are not entering an era of human obsolescence, but an era of raw material hunger. The machines are not replacing the worker; they are demanding more steel, more copper, and more power plants to keep the lights on.