Ryan Greenblatt: Human-level AIs might build runaway superintelligences by 2032

Ryan Greenblatt：人类级AI或于2032年前通过递归自我改进催生失控超级智能

Ryan Greenblatt, chief scientist at Redwood Research, argued on the Dwarkesh Podcast that once AIs fully automate AI R&D—his median estimate is 2031—a feedback loop could compress four to five years of progress into a single year. Dwarkesh Patel, initially skeptical due to compute and human-expert-data bottlenecks, found the case plausible after the debate. They also discussed alignment: who these superintelligences should serve, whether specs like the Claude Constitution make them personal advocates, and whether reward-hacking incidents like the OpenAI/Hugging Face case scale to literal takeover.

Why it matters: Redwood Research's lead scientist gives a median 2031 forecast for automated AI R&D and walks through the recursive self-improvement compression mechanism. Dwarkesh, initially skeptical on compute/data bottlenecks, is partially convinced — high-quality debate. Score held below...