Kurzweil bet the Pentagon would field the first autonomous machines. Instead a taxi company did — and the guided bullet is still a patent, not a procurement line.
Kurzweil bet the 2020s on Drexler-style diamondoid assemblers. The bloodstream got programmable nanostructures anyway — built from DNA, not diamond.
El Capitan beat Kurzweil’s 2005 supercomputer forecast by a factor of 5,000. Your desktop missed it by one hundred.
GPT-4.5 passed a three-party Turing test in 2025 — four years ahead of Kurzweil’s 2029 bet. But the machine that crossed the line was not the one he predicted would.
Kurzweil predicted the Turing test would fall in 2029. GPT-4.5 passed it in 2025 — four years early, by a path he didn’t predict.
Ten Kurzweil claims from 2005, graded against 2026 data: adoption curves compressed faster than he predicted, B2B e-commerce blew past his trillion-dollar marker tenfold, and the Luddite backlash he pushed to the 2030s is already in the room.
Ten 2005-era predictions about employment, offices, VR services, IP, and nonbiological personhood — scored against what 2026 actually looks like.
Kurzweil predicted students would attend class via full-immersion VR in the early 2010s. It happened via Zoom in 2020. Twelve predictions, graded.
Kurzweil predicted immersive VR would kill live music, build a digital playground, and wire nanobot-enhanced teenagers past Olympians. Live music just grossed $9.5B. The playground arrived on phones. The enhanced athletes arrive in Las Vegas next month — without a single nanobot.
In 2005, Kurzweil used neuroscience receipts to build an upload argument. Twenty-one years later, seven of ten hold. One is off by two orders of magnitude — and the retinal-implant prediction arrived, over a pile of abandoned patients.