Quantum computers, long heralded as the future of computational power, are edging closer to practical application, according to a report published in New Scientist. Recent breakthroughs by leading companies, including IBM, Microsoft, and Google DeepMind, suggest significant strides in overcoming longstanding challenges such as scalability and error correction. These advancements hint at a transformative impact across fields like materials science, chemistry, and beyond.
IBM showcased its latest development in quantum computing by successfully linking two of its Eagle quantum processing units, each containing 127 qubits. This innovative connection enabled the system to perform calculations requiring 142 qubits—exceeding the capacity of either individual chip. This was achieved using a method that involves entangling qubits and “teleporting” one to a second chip, mediated by a classical computer. Blake Johnson of IBM described this approach as a strategic way to scale quantum systems by breaking the problem into manageable chunks. While this marks an early step, experts like Scott Aaronson from the University of Texas at Austin see it as a foundational advance for building larger superconducting quantum systems.
Error correction, another critical hurdle in quantum computing, is also seeing notable progress. Microsoft and Atom Computing recently reported a record-breaking achievement by entangling 24 logical qubits, a crucial milestone for reducing computational errors. Their system, built using ultracold ytterbium atoms, achieved error rates four times lower than conventional qubit systems in early tests. Krysta Svore from Microsoft emphasized the importance of these logical qubits, saying they are essential for achieving breakthroughs in quantum applications. The team’s next goal is to create 50 logical qubits, with the ultimate aim of reaching 100, a threshold believed to unlock significant advancements in solving complex scientific problems.
Meanwhile, Google DeepMind has applied artificial intelligence to quantum error correction, developing a model called AlphaQubit. This AI-driven decoder leverages the same transformer neural network technology behind tools like ChatGPT and AlphaFold. By learning to interpret data from error-detecting qubits, AlphaQubit has achieved a 6% reduction in errors compared to existing algorithms. According to Johannes Bausch of Google DeepMind, this improvement, though seemingly modest, offers scalability that other methods lack, positioning it as a vital tool for future, larger-scale quantum systems.
Experts agree that these advancements collectively indicate a rapid maturation of quantum computing technology. Aaronson noted the increasing synergy between classical and quantum computing, particularly as machine learning tools like AlphaQubit enhance the efficiency and accuracy of quantum error correction.
