Musings on Recursive Self-Improvement - by Séb Krier
substack.com · 2,388 words · saved by 1 readers
Hard takeoffs are... hard
When thinking about recursive self-improvement, there are two things to separate out: whether we’re talking about models (and scaffolds) improving more rapidly, or the wider societal/economic impacts of this recursive dynamic - i.e., stuff that depends mostly on deployments, adoption, and other bottlenecks. The lines are sometimes blurred in the discourse, and I want to make this separation clearer here. I also suspect many of these barriers are more structurally resistant to being dissolved by better capabilities than people often assume. The models will get better On the former: yes, we…
saved by
related reading
- Data bottlenecks won’t prevent an intelligence explosionnewsletter.forethought.org
- When AI builds itself \ Anthropicanthropic.com
- Lossy self-improvement - by Nathan Lambertinterconnects.ai
- Why I think AI take-off is relatively slow - Marginal REVOLUTIONmarginalrevolution.com
- Why transformative artificial intelligence is really, really hard to achievethegradient.pub
- Import AI 455: AI systems are about to start building themselves.importai.substack.com
- Will AI R&D Automation Cause a Software Intelligence Explosion?forethought.org
- On Recursive Self-Improvement (Part I)substack.com
- Harness Engineering for Self-Improvement | Lil'Loglilianweng.github.io
- rsi-paper.pdfelasticity.institute
- Training Imperativesdan.io
- Does AI Progress Have a Speed Limit?—Asteriskasteriskmag.com