Quant²AI

Quantum computers are expected to offer immense advantages over classical computers, especially for resource-intensive applications such as artificial intelligence. The Institute for AI Safety and Security is investigating whether these hopes can be realised in the project Quantification of Quantum Advantages for Artificial Intelligence from a heuristic point of view. We will define an application-oriented benchmark for quantum AI pipelines by the end of 2025. This will help users to make a qualified decision for a method and to assess the expected benefits. To achieve this task, the institute for AI safety and security is joining forces with partners from Fraunhofer ITWM, Conet, Data Cybernetics and Jos Quantum in this project financed by the DLR Quantum Computing initiative.

The applications for artificial intelligence are constantly expanding, from autonomous systems to identifying customer needs. As a result, the demands on the models generated are increasing. However, optimising a model is a time-consuming process that requires a lot of data to be analysed using a lot of computing power. Many stakeholders expect the transition to quantum AI to improve the performance of the learning process. In many cases, these expectations are still not clearly formulated and are not supported by much evidence: For existing quantum AIs, there are mostly only feasibility studies, which consider only partial steps and are difficult to compare.

Quant²AI aims to enrich the discourse on possible advantages of different quantum AI approaches with quantitative insights. As a first step, we systematise possible benefits and map the performance of an AI system into a suitable metric. This will lead to a benchmark, a prototype of which should be available by the end of the project, together with a set of application-oriented example problems. This will cover the entire process from the preparation and coding of the data sets to the interpretation of the AI's results. This is necessary because gains from the use of quantum computers in individual sub-steps will only bring relevant benefits if they can also be incorporated into the AI pipeline without losing the benefits.

The datasets cover DLR's core areas such as aerospace, transport and energy. In addition, an industrial partner contributes further topics relevant to current applications. In the final step of the expansion, we will provide a number of data sets of varying complexity for different. Based on these data, we will evaluate the trained AI models in terms of metrics for expected quantum advantages.

Quant²AI gives users the opportunity to choose between different methods of quantum AI for their use case and to make a realistic assessment of the possible advantages over classical methods. Developers will have a tool for the evaluation of their algorithms and the identification of potential bottlenecks. Quant²AI thus provides an overview and comparability of quantum AI methods.