IBM Leverages AI to Accelerate Discovery of Quantum Error Correction Codes

IBM researchers have unveiled OpenEvolve, an open-source, LLM-guided evolutionary AI framework designed to accelerate the discovery of advanced Quantum Error Correction (QEC) codes. Finding optimal QEC codes is historically a computationally expensive task, but by utilizing large language models to hypothesize promising algebraic formulations, the new framework significantly streamlines the process.

The research team focused their testing on bivariate bicycle (BB) codes, a key class of quantum low-density parity check (qLDPC) codes identified in IBM’s fault-tolerant quantum computing roadmap. The evolutionary campaign successfully identified 465 new codes, offering diverse structural trade-offs. Notable examples include a design with 50 logical qubits—surpassing the previous record of 16 for this family—and hardware-optimized configurations that require only 72 physical qubits.

While further evaluation is necessary to test these codes in real-world environments, OpenEvolve provides a robust methodology for exploring complex algebraic code spaces. IBM has made the library available on GitHub to encourage collaborative development within the global quantum community.

Source: quantumcomputingreport.com
Publication date: 14.06.2026
Author: dougfinke