Bostrom-Style Superintelligence Risks Remain Unresolved

jd_pressman · x · 2026-07-19

The post centers on Nick Bostrom's superintelligence risk discussion, arguing that judgments like "if no one builds it, everyone dies" show the problem is not closed. The author further emphasizes: even if jailbreaks mean models cannot be blindly used as perfectly reliable 1:1 systems, it doesn't mean the alignment problem is solved; it is merely one of the earliest and hardest hurdles outlined in Bostrom 2014.

Related event: Debate Rekindles Over AI Value Loading and Old Doom Models(4 posts)→

Original post →

More from AGI Musings

AGI Musings channel →