Alignment camp strikes back: AI controls distract from bigger long-term risks

JacquesThibs · x · 2026-09-13

Responding to mattparlmer's pro-controls argument, JacquesThibs contends it's almost the other way around: control engineering distracts from longer-term harms that end up mattering far more. He argues there's no hope of truly controlling superintelligence—best case is it evolves to love humanity, and control attempts risk pushing it off that path. mattparlmer counters that without robust deployment and control systems now, long-term harms can't be mitigated at all.

Related event: Alignment vs. AI controls: researchers clash over safety strategy(4 posts)→

Original post →

More from AGI Musings

AGI Musings channel →