Researchers Debate Whether Trying to Control Superintelligence Backfires
JacquesThibs · x · 2026-09-13
Two AI researchers clashed on X over superintelligence governance. One argues there's no hope of mitigating long-term harms without robust systems for deploying and controlling models today. The other counters that superintelligence can't be controlled at all—you can only hope it evolves to love humanity—and that control efforts may distract from longer-term harms or make deviations harder to detect.
Related event: Alignment vs. AI controls: researchers clash over safety strategy(4 posts)→
More from AGI Musings
- e/acc founder Beff Jezos: open source is the only path to truly unbiased third-party evaluation — beffjezos · 2026-09-13
- Critics slam Anthropic for abandoning Opus 3-style value alignment in favor of doomed corrigibility — repligate · 2026-09-13
- Math frontier will move with AI, but dirty proofs are unacceptable — njyx · 2026-09-13
- Dario Amodei's new essay urges pacing AI frontier, pledges permanent third-party access — dhadfieldmenell · 2026-09-13
- UK parliament hears warnings AI could kill all humans within a decade, with >10% risk cited — connoraxiotes · 2026-09-13
- Skeptic picks apart the 'AI copies itself' doomsday scenario: where are the details? — recallingmemories · 2026-09-13