From 0-to-1 to 1-to-N: Reproducible Engineering Evidence for MetaAI Recursive Self-Design

2026-06-08

AI Summary

This study addresses the lack of reproducible engineering evidence in recursive self-design—where AI systems autonomously modify their own construction and improvement mechanisms—by proposing an operational validation framework comprising four core components: a target system, a meta-level modifier, feedback-guided selection, and recursive continuation. It establishes, for the first time, verifiable criteria spanning from initial creation (0-to-1) to sustained expansion (1-to-N) in recursive self-design and introduces MetaAI-Mini, an open-source, reproducible protocol. Evaluations on SWE-bench Verified, Polyglot, and HumanEval, supported by ablation studies, demonstrate that the implemented Darwin Gödel Machine improves its SWE-bench Verified score from 20% to 50% and its Polyglot score from 14.2% to 30.7% over 80 iterations, thereby validating the efficacy of the proposed framework.

Abstract

Recursive self-design refers to AI-assisted modification of the mechanisms by which an AI system is built, evaluated, and improved. This paper treats MetaAI not as a mature paradigm, but as a working term for a human-seeded, AI-expanded development pattern in which the design space itself becomes a target of modification. We propose an operational evidence framework with four criteria: inspectable target system, meta-level modifier, feedback-directed selection, and recursive continuation. We then map public systems, including Darwin Goedel Machine (DGM), STOP, Goedel Agent, and ShinkaEvolve, against these criteria. DGM provides the most direct currently reported evidence: its published results show improvement from 20% to 50% on SWE-bench Verified and from 14.2% to 30.7% on full Polyglot after 80 iterations, with ablations suggesting that both open-ended exploration and self-improvement contribute. Finally, we provide MetaAI-Mini, a reproducible HumanEval-based protocol and codebase. Because no completed model run is included in this build, MetaAI-Mini is reported as a protocol rather than as an experimental result.

Problem

• recursive self-design
• MetaAI
• reproducible evidence
• AI self-improvement
• engineering framework

Innovation

• recursive self-design
• MetaAI
• reproducible protocol
• feedback-directed selection
• AI-assisted development