On October 22, 2025, Google Quantum AI published a paper demonstrating the ability to efficiently probe the dynamics of highly entangled many-body systems—something that classical computers have struggled with. They achieved this by using time-reversal techniques, similar to the Loschmidt echo in NMR systems, combined with Randomized Benchmarking. While this isn’t directly related to my own work, the techniques they developed are quite impressive.

What is particularly interesting to me is their accompanying blog post, titled “A Verifiable Quantum Advantage,” which discusses the importance of verification in quantum computing.

In classical computing, we can always check the result of a computation using smaller, trusted devices, or ultimately by hand. But in the quantum world, this is impossible because a quantum advantage inherently means a situation where classical computers cannot handle the problem. If we didn’t find a way to verify the result, it would imply that we would have to blindly trust the quantum computer. Fortunately, this isn’t the case.

Google suggests a simple method for verification: because the task is to estimate the expectation value of an observable, and not a random sample, the same computation can be repeated on a different quantum computer, with the expectation that the results should match. While this is a significant step forward, it oversimplifies the challenge.

The reason is that quantum circuits on different machines are not identical. For example, Google’s circuits are specifically tailored with error mitigation strategies that are unique to its hardware. Running the same circuit on a different chip, which has a different error mitigation approach, will likely produce slightly different results. So, the question remains: how close do the results need to be in order to claim reproducibility?

This is where more advanced verification techniques come in. One such technique is detailed in our paper, Verifiable Blind Observable Estimation: A Composably Secure Protocol for Near-Term Quantum Advantage Tasks, which I co-authored with Bo Yang and Elham Kashefi. While Google’s machines can’t fully implement these protocols yet—due to their need for mid-circuit measurements—our approach provides a concrete way to guarantee the accuracy of quantum computation results. By repeating computations and using advanced error-handling techniques, we can provide a confidence level for the results, ensuring that the quantum advantage can be verified in a robust and reproducible way.

This is an exciting time for the quantum community, and it’s encouraging to see verification gaining more attention. We still have a long way to go, but this work marks a significant milestone in establishing trust in quantum computation.