Steering Committee of Plattform Lernende Systeme: exploiting the potential of AI

Munich, 31 January 2023

The text-based dialogue format ChatGPT is a further milestone in the development of artificial intelligence (AI). The members of the Steering Committee of Plattform Lernende Systeme (PLS) discussed with guests at its eleventh meeting the opportunities and challenges presented by technological progress in major linguistic models. Other topics of discussion included: What can we expect from hybrid AI? How can business data be safely used to add value? And how does AI change journalism?

The hybrid meeting co-chaired by Federal Minister of Research Bettina Stark-Watzinger and Reinhard Ploss took place at the end of January in Berlin and online. On the agenda were recent developments in the area of AI, on which the working groups of Plattform Lernende Systeme are currently growing expertise.

Volker Tresp,  a professor at Ludwig-Maximilians-Universität (LMU) Munich and head of the PLS working group “Technological Enablers and Data Science” gave a summary of the performance of major linguistic models. He outlined possible business applications of ChatGPT, such as digital assistants for customer service or technical support. In the subsequent discussion, the need for a German linguistic model was pointed out. In her presentation, Elsa Kirchner highlighted the advantages we can expect from linking data- and rule-based AI systems – so-called hybrid AI. A professor at University of Duisburg-Essen, she leads the PLS working group “Learning Robotic Systems”. Hybrid AI leverages existing knowledge and can be developed using less resources than purely data-driven AI applications. The challenge is the optimum interaction of both approaches as well as the availability and transferability of data to other systems, said Kirchner.

Facilitating data use, adding value

Jörn Müller Quade, a professor at Karlsruhe Institute of Technology (KIT) and head of the PLS working group “IT Security, Privacy” presented technical approaches for the use of data for AI systems in compliance with data protection. He made the case for enhancing the legal certainty for companies and introducing flexibility into the use of AI data towards the common good. Jürgen Müller, CTO at SAP and member of the PLS Steering Committee outlined how AI can contribute to added value in various areas of the company. He illustrated this based on a use case for AI-assisted planning forecasting for a food company. Uli Köppen’s presentation centred on the impact of AI on the working environment. Head of the AI and Automation Lab of the public-service radio and television broadcaster Bayerischer Rundfunk, she described how the broadcaster uses AI for researching, producing and distributing articles. PLS recently published a white paper on AI in journalism (in German).

After a short retrospective from Birgit Obermeier (acatech Office) on the 2022 Digital Summit and further communication highlights from Plattform Lernende Systeme, Wolfgang Wahlster (DFKI), who is a member of the Steering Committee, presented the initiative AI Grid, which is funded by the Federal Ministry of Education and Research (BMBF). The aim of AI Grid is to support young AI talent in doing research for their Master’s or doctoral thesis and help build up an international network taking a grassroots approach.

The importance of the trustworthiness of AI applications is something that keeps coming up in the discussion on the individual topics of the meeting. Enhancing trustworthiness remains on Plattform Lernende Systeme’s agenda.