Ilya Sutskever Hints SSI May Pivot from 'Straight-Shot' ASI to Incremental Deployment

kiki-le-koala · reddit · 2026-08-05

In an interview with Dwarkesh Patel, Ilya Sutskever discussed Safe Superintelligence (SSI) strategy, suggesting they might not stay in stealth until ASI but deploy models incrementally. Reasons: people need to see AI in action to grasp its power; if ASI takes 5-20 years, staying closed is unrealistic; real-world feedback is crucial for safety.

Original post →

More from AGI Musings

AGI Musings channel →