Despite all the rapid progress in the field of artificial intelligence (AI), more and more serious problems with computing, i.e. IT infrastructures and networked systems, have become known worldwide in recent years. These problems may severely restrict the further development of AI and future technologies - especially communications, medicine and robotics - or, in the event of a problem with energy supply to the systems, bring them to a standstill. This is because AI applications consume enormous amounts of energy.

Renowned researchers from TUD Dresden University of Technology, the Ludwig-Maximilians-Universität München (LMU) and the Technical University of Munich (TUM) will now tackle this topic together as part of the "GAIn" (Next Generation Al Computing) pilot project. Prof. Frank Fitzek, holder of the Deutsche Telekom Chair of Communication Networks and spokesperson for the Cluster of Excellence Centre for Tactile Internet with Human-in-the-Loop (CeTI), and Prof. Stefanie Speidel, Chair of Translational Surgical Oncology at the National Center for Tumor Diseases Dresden (NCT/UCC) and CeTI spokesperson (both TUD), will be participating on the Saxon side. Prof. Holger Boche (TUM) and Prof. Gitta Kutyniok (LMU) will be involved in the project from the Bavarian side.

The Free States of Saxony and Bavaria will be providing EUR six million until 2027 for transnational scientific cooperation in the GAIn pilot project. TU Dresden has earmarked EUR three million for the project between 2024 and 2027. The first allocation of around EUR 500,000 has already been made by the Saxon Ministry of Science.

Prof. Ursula Staudinger, Rector of TUD: "The GAIn project is based on the globally recognized expertise of the University of Excellence TUD in the interaction between hardware and AI applications, builds on the CeTI Cluster of Excellence and strengthens our existing cooperation with the Munich University Excellence Network."

Prof. Frank Fitzek (TUD): "In my view, the GAIn project offers a great opportunity to significantly shape AI-operated communication networks. Thanks to our research and expertise in future communication networks, we can actively contribute to solving challenges in energy consumption, computing power and reliability of AI systems and create added value for society at the same time. The transnational cooperation between Saxony and Bavaria will not only strengthen Germany's technological sovereignty, but will also be important for the CeTI excellence application."

Prof. Stefanie Speidel (TUD): "The development of new AI hardware and software concepts as part of the GAIn project is of great importance to me to enable further progress in medical technology AI and robotics research. By reducing the energy consumption and increasing the stability of AI systems, we can ensure that assistance systems in surgery work even more precisely and efficiently. These innovations mean that surgical procedures can be performed more safely and in a more targeted manner, which ultimately benefits patients."

Saxony Minister of Science Sebastian Gemkow: "By launching the GAIn research project, we want to give Saxony and Bavaria an international leadership role in central computing technologies and thus also contribute to Germany's technological sovereignty. Saxony has gathered impressive expertise in the fields of visionary hardware, communications and robotics and is a European leader in microelectronics thanks to the establishment of large chip factories and innovative start-ups. The collaboration between Saxony and Bavaria promises to combine the outstanding research and innovation expertise available in both states and to work together on completely new AI hardware and corresponding software concepts. I am delighted that we are able to support this outstanding project in Saxony with three million euros."

Bavarian Minister of Science Markus Blume: “Together we are building the path to a new age: Where there are still frontiers today, Bavaria and Saxony are expanding the horizons of AI research. With ever-increasing demands in medicine, robotics and communications, our technological progress must also increase in scale. Energy-efficient hardware and pioneering software concepts will be our key to achieving these goals. Even if we cannot imagine today what will be possible in the future, we must create a foundation now that we can build on in the future. Our two Munich Universities of Excellence and TU Dresden will be an excellent team for this purpose - we will be happy to support them in Bavaria with EUR three million over the next three years. In this way, we are using the unique boost of the EUR 5.5 billion High-Tech Agenda and the new Bavarian AI offensive to provide further international impetus and strengthen Germany's strategic position in this crucial future field in the long term."

As part of "GAIn," the development of innovative AI hardware and corresponding software concepts is intended to solve problems in the areas of energy consumption, predictability, reliability and legal implementation. The idea is to significantly reduce the energy consumption of AI-based applications and achieve predictability and reliability. AI technologies should also be able to meet legal requirements.

Background: Challenges in the development of AI hardware and applicationsThe

further development of AI hardware and applications is facing challenges in the areas of energy consumption, predictability, reliability and the implementation of legal requirements (such as the EU AI and EU Data Act).

- Computability: AI solutions to many problems are still not computable on current hardware platforms.

- Reliability of AI: AI applications are not yet reliable in many respects, as shown by the unexpectedly slow further development of autonomous driving despite massive investments by large and well-known companies, among other things. The hardware used to date (CPUs/GPUs) has proven to be a causal problem in scientific studies.

- Legal issues: AI applications trained on current hardware platforms cannot fulfill the legally required "algorithmic transparency" and the "right to explanation" for various critical classes of problems.