He Discusses Second-Order Alignment in Human-Machine Swarms

anderssandberg · x · 2026-07-14

In the upcoming book *Law Liberty Leviathan*, Sandberg discusses a "second-order alignment problem": cognitive synergy among humans, machines, and human-machine hybrid groups remains a largely unexplored domain. He echoes a previous point: rather than merely focusing on whether individual models are aligned, we should be more wary of the complex alignment issues that emerge from the interaction of humans, machines, and institutions.

Related event: Sandberg Warns of Useful AI Transition Shocks and Human-Machine Alignment Gaps(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →