Redwood Researcher Predicts AI Takeover by 2029: How the Transition Unfolds
mattturck · x · 2026-08-28
Matt Turck interviews Ryan Greenblatt from Redwood AI about the future of AI, alignment risks, and the transition to superintelligence.
- CEO Awareness: AI CEOs are aware of the risks but are "proceeding anyway."
- Nature of Threat: Superintelligence is described not just as "bad," but as "dangerous."
- Recursive Self-Improvement: They discuss technical realities vs. intuition objections.
- SSI Rumors: How continual learning might impact the landscape.
- Timeline: Ryan suggests planning as if it happens in 2029 and debates if it's already too late.
- Background: Ryan's path from COVID to Redwood, including the recent "alignment faking" story.
Related event: Redwood Researcher Predicts AI Takeover Could Come by 2029(3 posts)→
More from AGI Musings
- If Superintelligence is Aligned to Preference Utilitarianism, Human Labor Becomes Answering Bizarre Ethics Experiments — moultano · 2026-08-29
- Essay: Data centers are next-generation ports — communities will build around the flow of intelligence — McDonaghMatthew · 2026-08-29
- Humans may eventually become motivational coaches for AI while AI does all the work — XFreeze · 2026-08-29
- AI Trends Are Often Simplifications of Existing Tech, Not Breakthroughs — eptwts · 2026-08-29
- Study shows AI bias: people reject human answers thought to be AI — VraserX · 2026-08-29
- If AI Hasn’t Changed Your Life, You’re Probably Using It Wrong — YogeshMalik · 2026-08-29