Another End-of-the-World Bet

With Greg Colburn

In 2017, I made an end-of-the-world bet with Eliezer Yudkowsky. Despite knee-jerk incredulity, betting on the end of the world is easy. Key mechanic: The optimist prepays the pessimist. If the world ends, the pessimist has extra spending money during the Last Days. If the world doesn’t end, the optimist gets his money back, plus more. The central term of the Caplan-Yudkowsky bet:

If there are still biological humans running around on the surface of the Earth [on January 1, 2030], it will not have been said to be ended.

For Eliezer, as you’ve probably heard, AI is the expected agent of destruction. And while he’s got legions of disciples, no one else has tried to make the same bet with me.

Until now.

The brave Greg Colburn has just accepted the same terms as Eliezer did back in 2017. I’ve paid him $250, and if the world doesn’t end by January 1, 2030, Colburn will repay me $500 inflation-adjusted.

Since I scoff at the prospect of AI-driven human extinction (Bah!), this is a much better deal for me than the original bet. Instead of doubling my money in 12 years, I’ll double it in less than 4 years. But since AI development since 2017 has been much faster than most people expected, AI doomers should have the opposite opinion: My defeat is even more foreseeable in 2026 than it was in 2017.

P.S. I’ve never met an AI doomer who claimed to be massively borrowing in order to maximally enjoy their final years. But if doomers took their pessimistic view seriously, they obviously would so borrow.