Most Likely Cause of an Apocalypse on December 21

Note: This post is al­most com­pletely tongue-in-cheek. Ob­vi­ously the chances of De­cem­ber 21, 2012 herald­ing in an apoc­a­lypse, defin­able maybe as an event caus­ing billions of deaths and/​or global catas­trophic in­fras­truc­ture dam­age, are slim to none.

But they aren’t ac­tu­ally none.

Let’s say there’s a 5% chance of a su­per­hu­man Gen­eral AI be­ing de­vel­oped in the next 10 years and ush­er­ing in the sin­gu­lar­ity. Let’s say 45 of those sce­nar­ios would lead to a Bad End which could rea­son­ably be called an apoc­a­lypse (or an “AI-poca­lypse”, per­haps). And let’s say the dis­tri­bu­tion of prob­a­bil­ity isn’t lin­ear, but is ex­po­nen­tially skewed some­how so that given the emer­gence of AI in the next 120 months, the chances of it hap­pen­ing in this very month are 1 in 100,000. Then let’s di­vide that by 30 so we can get our apoc­a­lypse rol­ling on the right day.

5100 * 45 * 1/​100000 * 130 = 1 in 75,000,000

Ad­mit­tedly, low odds. This com­ports with the fact that there would have to be a lot of un­known de­vel­op­ment progress be­ing made already on the prob­lem for an in­tel­li­gence ex­plo­sion to be any­where on the hori­zon.

But com­pared to some of the sce­nar­ios de­bunked by NASA (http://​​www.space.com/​​18678-2012-mayan-apoc­a­lypse-fears-nasa.html), such as a col­li­sion with the rogue planet Nibaru, or Earth be­ing sucked into the su­per­mas­sive black hole in the cen­ter of the Milky Way 30,000 light years away, the AI-dooms­day sce­nario starts to seem rel­a­tively plau­si­ble.

I think the only other (rel­a­tively) plau­si­ble con­tenders would be the re­lease of a pan­demic-caus­ing biolog­i­cal weapon, or the start of an in­ter­na­tional nu­clear war. I haven’t done any Fermi calcu­la­tions on those, but I’m sure their prob­a­bil­ity ex­ceeds that of so­lar flares scourg­ing the sur­face of the Earth.