Google cuts quantum computer errors by 40-fold
The Google Quantum AI team demonstrated on the Willow processor the "magic state cultivation" technique, achieving a logical fidelity of 99.99% and opening, according to the company itself, the way for further developments needed to scale the method.
Google Quantum AI announced on October 16 a breakthrough in quantum computer error correction, demonstrated on the Willow processor through a technique called “magic state cultivation”. The result, published in the journal Science, reduces errors by a factor of 40 compared to non-cultivated states, achieving a logical fidelity of 99.99% for T states.
The research team that carried out the work includes Craig Gidney and Noah Shutty, the same researchers who had developed the original method in 2024. The transition from theory to practical demonstration on the Willow processor therefore represents the continuation of work already begun two years earlier by the same authors.
The issue of errors remains central to any discussion of quantum computers. As science journalist Michelle Starr summarized, “quantum computers have an awkward problem: they’re extraordinarily good at making mistakes” (transl. from English), a description that frames why error correction is considered one of the main obstacles to the development of reliable quantum machines.
Despite the demonstrated improvement, Google itself states that further developments are needed to scale the technique beyond the experimental context in which it was demonstrated. The transition from a fidelity demonstration on T states to a fully operational quantum system on a large scale therefore remains a future goal, not an already achieved milestone.
The figure of 99.99% logical fidelity and the 40-fold reduction in errors currently remain the documented experimental result: publication in a peer-reviewed journal such as Science places the work within the field of verifiable research, while the same company that conducted it acknowledges the distance separating this demonstration from an industrial-scale application.
Sources: ScienceAlert; The Block
← Archive · Front page · Past editorials · Report an error · Original article (in Italian)