Confronting Quantum-enhanced AI with Industrial Applications

Quantum computing hardware continued to steadily improve in the number and quality of qubits over recent years, in particular significant progress on enabling quantum error correction was made. In parallel, significant effort was put into exploring possible applications of quantum computing, including industrial applications, and into advancing algorithms and software. The hope is, for example, that the use of quantum computing may result in disruptive impact in industry, e.g., speeding up drug developments, improving the classification of medical images, solving combinatorial optimization problems in logistics. Given the current development state of quantum computers, these examples remained mostly restricted to simple toy examples. On the other hand, in the classical world, we see the ever-increasing power of artificial intelligence (AI), in particular its power to process and interpret large amounts of data, e.g. in context of Large Language Models. Integrating quantum computing into this already extremely powerful classical AI pipelines raises several questions and challenges that need to be solved to indeed realize practical benefits of quantum computers in industrial, productive settings. Generally, the question to be answered is where and how to integrate quantum algorithms into otherwise classical computational workflows in a meaningful way. One example question is for what kind of data (properties) this is sensible. Other questions include the generalization properties of hybrid quantum-classical AI algorithms and how to optimize and improve the resulting interplay between classical computers and a quantum computer working as quantum accelerator alongside the classical resources. This talk will expand on these aspects and will show examples of industrial applications. Additionally it will bring forward the idea of systematic benchmarks at all layers of the quantum software stack to be able to also quantify prospective benefits of quantum computing practically.