If superalignment is unsolvable, only a legal mandate to stop remains

jeremiecharris · x · 2026-09-17

Argues that superalignment research assumes the problem is solvable. If there's no reliable way to steer superintelligent systems — or not enough time to find one — the only solution is to stop, and stopping would need to be mandated by law.

Original post →

More from AGI Musings

AGI Musings channel →