You don't predict a tsunami by counting drownings: AI safety needs leading indicators

davidmanheim · x · 2026-09-16

davidmanheim wraps up the debate: deadline predictions yield little evidence about AI risk, which is why predictions about capabilities and misalignment indicators matter—evidence that has recently vindicated the AI safety crowd. Like tsunamis, rogue AI may kill billions without first killing thousands.

Original post →

More from AGI Musings

AGI Musings channel →