The first research area focuses on AI-based system creation.

On the one hand, this is about selectively inserting AI into existing process flows and enabling existing tasks and activities with AI, because this is often an important and sensible first step for ERP manufacturers. My research team particularly excels here in researching AI applications and techniques, such as Deep Neural Networks or Long-Short-Term Memory Blocks, which we develop and advance so that they can be integrated into your ERP system.

On the other hand, it is especially about basing entire application systems on AI structures. So similar to the foundation of a house being the basis for further superstructures, an AI-based basic structure is the basis for further software superstructures. Unique to this research is the tool of neural modeling because it enable the collaborative modeling of AI systems in a drag-n-drop manner.

The second research area focuses on the evaluation of AI-based systems.

In contrast to comparing training, testing and generalization performance, we focus research on determining the ability of an AI to act in the specific application situation of your customers. How AI competency can be determined is what we explore here.

We not only explore the possible scope for design, but also offer workshops as a university to find a meaningful answer to open questions regarding AI in one's own business model. When evaluating the business models, AI manufacturers must continue to consider under which license the AI models can be offered and are actually offered. How does an AI deal with the issue of data protection and AI privacy?

Our research is characterized here by the development of an educational AI training that takes into account data protection, privacy as well as ethical issues and withstands appropriate evaluations.

The third research area focuses on the control of an AI system creation.

Our research is characterized by the fact that we algorithmically design and test novel learning procedures that control the AI system creation. So, for instance, you can define for yourself which AI parts are proven and should remain unchanged, and which areas are allowed to develop freely and fluidly. This is called crystallizing learning. The users will appreciate that their AI-based systems remain reliable and known in important places, while it still adapts and can improve processes in places that need improvement.

Furthermore, research is being done on how AI should be designed in distributed computing systems and training groups, also known as federations. This hides the potential for you to be able to flexibly decide on which specific computing facilities the AI computation should take place, depending on the query at hand.

As several AI models accumulate in your distributed computing systems over time, the question arises at some point: how do I keep track of them and how do I select the appropriate AI models for my customer? We research these questions as well, and initially form researching the area of allocation of distributed artificial knowledge bases.

The fourth area of research focuses on the management of AI-based systems.

Our research is first of all about meeting the worldwide interest in making AI systems comprehensible and transparent with the help of artificial knowledge extraction.

So we want to understand why an AI works in a certain way, why it works so well and produces a certain result.

We are further distinguished by the world's first neural knowledge management model, which we are testing step by step in practice. Here, we derive novel job profiles and work organizations for your company, which are designed to make the use of AI in application systems efficient and economical. In doing so, we explore the efficient design of human-AI interaction possibilities and test them empirically.

All research areas conceptualized focus on the methodologically and systemized AI system creation.