Ryan Greenblatt – What happens once AI can automate AI research?
dwarkesh.com · 10,627 words · saved by 1 readers
A debate about recursive self-improvement.
Had Ryan Greenblatt on to discuss/debate recursive self-improvement. This might be the most important question in the world right now – whether within a year or so of achieving human-level intelligence, you slingshot towards having 10s of billions of superintelligences, each of which is dramatically more competent than human experts across all fields. I’ve historically been skeptical of this possibility. My intuition has been that we will end up significantly bottlenecked by not only compute scaling but human expert data, which I think underlies most of the AI progress today. If, because…
saved by
related reading
- Will scaling work?dwarkeshpatel.com
- When AI builds itself \ Anthropicanthropic.com
- AI researchers debate how close we are to recursive self-improvementyoutube.com
- AI 2027ai-2027.com
- OpenAI researcher on agent swarms & recursive self-improvementyoutube.com
- My picture of the present in AI - by Ryan Greenblattblog.redwoodresearch.org
- Import AI 455: AI systems are about to start building themselves.importai.substack.com
- My picture of the present in AI — LessWronglesswrong.com
- Unresolved debates about the future of AIhelentoner.substack.com
- Does AI Progress Have a Speed Limit?—Asteriskasteriskmag.com
- Andrej Karpathy — AGI is still a decade awaydwarkesh.com
- Lossy self-improvement - by Nathan Lambertinterconnects.ai