AI safety camp clashes with prominent rationalist over 'alignment is really hard' prediction

JMannhart · x · 2026-09-06

A public spat in AI safety circles: the author mocks a prominent scholar for banning replies and publicly rejecting the 'alignment is really hard' prediction without stating what evidence would change his mind. The quoted tweet ridicules the scholar's 'we program them, so they'll obey' take as reflecting at most five minutes with the Wikipedia page on the alignment problem.

Related event: Pinker slammed for dismissive take on AI alignment(4 posts)→

Original post →

More from AGI Musings

AGI Musings channel →