IonQ, a Maryland-based quantum computing company, engaged in the second Quantum Natural Language Processing conference in Oxford, where they described some of their current language processing systems among other researchers. The last QNLP conference was held in 2019 when quantum computing programming was in its infancy. The first successful NLP implementation on quantum hardware was in 2020. New ideas and results from AI research using quantum computers have been published more frequently. Real quantum implementations are currently becoming more prevalent and rapidly! NLP is applied to image categorization, financial modeling, and probabilistic reasoning.
Quantum mechanics and natural language processing use the same algorithm for fundamental operations, and IonQ has been creating prototypes of several NLP components. The first QNLP application took the industry more than ten years to develop, but IonQ, In just a few months, was able to build the second, third, and fourth using IonQ Harmony. At the conference, IonQ used the NLP technique or language theory examples that can be helpful in various AI fields. A common theme in all their examples is the potential to combine multiple components and use the final product to execute computational tasks.
The most important finding is that they can create quantum versions of conventional AI methods using the common mathematical language of vectors. The takeaway from their illustration is that quantum computers perform actual language processing tasks. But they’re still at the early stage. 210 variables, or a conventional kilobyte, could be represented by an earlier ten qubit system. However, the modern IonQ Forte has thirty or more qubits, theoretically allowing them to define 230 variables or a gigabyte. However, the systems mentioned above do not yet use this scale since it’s unclear how to achieve this optimally.