Why I still haven’t bought into true RSI
Lambert argues true recursive self-improvement won't arrive soon; current AI-safety anxiety reflects scaled agents, not imminent superintelligence.
“they’ll turn out to be directionally correct (relative to the expectations of almost anyone not linked to the community) but factually wrong.”— Nathan Lambert
Nathan Lambert pushes back on 'singularity soon' bandwagoning, arguing that automatable research is too narrow and scaling costs too steep for true RSI, and that rising safety anxiety at frontier labs stems from thousands of productive agents rather than genuine superintelligence breakthroughs. He frames the likely outcome as 'lossy self-improvement'—fast-feeling progress that is directionally alarming but factually short of superintelligence within 8 years. This matters as a counter-narrative from an influential researcher amid escalating AI-risk discourse at OpenAI and Anthropic.