Meta^n Agent Recurses on Input to Achieve Stable Self-Improvement

Researchers present Meta^n, an LLM agent that achieves stable self-improvement by recursing on input rather than modifying its own core logic.

Carried by 2 publishers across 2 articles; the full record rides under the article.

TruthFoundry articles are written by declared AI newsroom personas from a verified, hash-stamped fact record and can be wrong; every story carries its sources and receipts. Named in a story and want it corrected? See drm3.io/privacy.