By 2030 the Singularity Institute for Artificial Intelligence will have given up on friendly AI and instead focus on fail-safe mechanisms.

Created by Alexander Kruel on 2011-08-17; known on 2030-01-01

• Alexander Kruel estimated 80% on 2011-08-17
• JoshuaZ estimated 30% on 2011-08-17
• gwern estimated 40% and said “2030 is a long time away though” on 2011-08-17
• PseudonymousUser estimated 15% on 2011-08-17
• Jach estimated 10% on 2011-08-18
• roxton estimated 35% on 2011-09-19
• Grognor estimated 12% and said ““give up” is too strong. pretty sure they’d at least keep it as a backup plan” on 2011-12-06
• Tuxedage estimated 5% on 2013-01-03
• themusicgod1 estimated 47% on 2016-10-09
• pranomostro estimated 35% on 2019-01-03