Seth Lazar: We're at the Threshold of 'Whack-a-Mole AGI'—and Must Decide Whether to Build It

sethlazar · x · 2026-10-01

In a talk at Johns Hopkins' Bloomberg Center, AI scholar Seth Lazar argued we're at least at the threshold of 'whack-a-mole AGI': any domain brought into distribution with a good judge can be made superhuman, enough to call AGI near. His second sense of reckoning: humanity, especially liberal democracies, must actually answer 'should we build AGI?'—AI is profoundly physical, so we could stop if we chose to.

Original post →

More from AGI Musings

AGI Musings channel →