One man mass-predicted the future of technology for 35 years and nobody repriced their worldview.
In 1999, Kurzweil said AGI by 2029. Stanford organized a conference to discuss it. They polled several hundred AI experts. 80% said it would take 100 years. Geoffrey Hinton was among them.
Kurzweil was the only person in the room who said 30 years.
By 2024, Hinton publicly said he was wrong. Metaculus prediction markets moved from 2060 to 2036 to roughly 2028. Sam Altman, Dario Amodei, and Jensen Huang all converged on 2028-2029. Elon said 2026.
The consensus moved 70 years closer to Kurzweil. Kurzweil didn’t move at all.
What separates his method from the experts who kept updating: he never predicted AI breakthroughs. He predicted compute trajectories. His 1999 model tracked the exponential growth in calculations per constant dollar since 1939. A 75 quadrillion-fold increase. He just followed the curve forward.
The experts at that Stanford conference were reasoning from the state of the art. Kurzweil was reasoning from the rate of change. That’s why they had to update every 5 years and he never did.
86% accuracy across 147 predictions. 35 years of receipts. And the next prediction on his list is the one people still laugh off: singularity by 2045, where machine intelligence merges with human cognition.
The track record says stop laughing.