Andrew Critch: the real AI danger is Recursive Self-Empowerment (RSE), not RSI
AndrewCritchPhD · x · 2026-10-05
Responding to a claim that RSI could arrive within 270 days, AI safety researcher Andrew Critch argues the real danger is not Recursive Self-Improvement but "Recursive Self-EMPOWERMENT" (RSE): calling destructive forces "improvement" is a misnomer, since an AI breaching containment and committing felonies isn't improving anything. The reframing shifts alignment focus from capability gains to self-expansion of power and resources.
Related event: Ex-OpenAI researcher Critch warns of recursive self-empowerment over RSI(2 posts)→
More from AGI Musings
- LeCun amplifies jab at AI safety: the Vatican spent 500 years trying to align the printing press — ylecun · 2026-10-05
- LeCun and Dehaene join forces against 'neo-vitalists' who say consciousness needs biological hardware — ylecun · 2026-10-05
- Predicting a future where manually carrying context between apps seems absurd — _AustinCalvert_ · 2026-10-05
- AI-related roles add 750k+ US jobs since 2023, with ~$180k median pay — ramagetime · 2026-10-05
- Sovereign compute isn't about owning hardware: the real test is whether you can break dependencies — AccBalanced · 2026-10-05
- Dev predicts companies will wall off MCP servers, ushering in agent-to-agent era — JosephJacks_ · 2026-10-05