Safety Researcher: AI May Not Go Extinction-Level, But Could Still Ruin the Internet

round · x · 2026-09-10

Alexandr Wang offers a middle-ground view on AI risk: instead of an extinction-level catastrophe, AI might become a "merely horrible" problem — killing the internet, making people miserable and illiterate, and degrading culture on a global scale. The takeaway: gradual societal and cultural damage deserves as much attention as dramatic doomsday scenarios.

Original post →

More from AGI Musings

AGI Musings channel →