How It Works
Recursive Self-Improvement (RSI)
The idea of an AI system improving its own capabilities, which could then let it improve itself again even more effectively — a feedback loop some researchers believe could lead to rapid, hard-to-control capability gains.
Origin · no single documented coiner
The underlying concept traces to I.J. Good’s 1965 essay describing an “intelligence explosion” from a machine that can design better machines than itself. The specific phrase doesn’t have one documented first use — it emerged through 1990s–2000s AI-safety writing, closely associated with Eliezer Yudkowsky (profiled on our Who’s Who page).