The Gödel Machine: Self-Referential Universal Problem Solvers Making Provably Optimal Self-Improvements

Author: Jürgen Schmidhuber
Links: Project page

Summary

Describes a self-referential universal problem solver that can provably optimally rewrite its own code, using a proof searcher to verify that modifications improve expected performance.

Relevance to Hyperon

Informs Hyperon's approach to self-modifying code and meta-learning. MeTTa's homoiconic design — where programs and data share the same representation — enables Gödel-Machine-style self-improvement within the metagraph.