Definition
Plain language
A system that only feeds on its own outputs, with no fresh input from outside.
As stated in the literature
A self-referential generative regime where each step's input is the previous step's output, with no exogenous data; subject to information-theoretic contraction by the data processing inequality.
Also called: closed-loop
Why it matters: Closed-loop generative training tends to drift and collapse, which is a growing concern as the web fills with AI-produced content.
For example, a language model is repeatedly trained only on text generated by previous versions of itself, with no fresh human-written data added.
Heard on the show
“… And the part the paper gestures at but doesn't quite earn — the closed-loop self-improvement vision where one capable model plans, executes, judges, generates curricula, …”Episode 008 — Why Long-Horizon AI Agents Get Stuck, and a Milestone-Based Fix That Helps