Increasing trust in artificial intelligence

With his ERC Starting Grant, Nils Jansen is researching how artificial intelligence applications can become more reliable and safe.

Artificial intelligence (AI) is already part of everyday life and will increasingly find its way into more areas of life. Applications in healthcare, transport and the financial sector are just a few examples. However, all of these areas require reliable and safe applications of AI systems. With the technology of reinforcement learning, AI applications can learn to make autonomous decisions for optimal results. So far, however, this has been done by trial and error, which basically means that all possible solutions are tried out until a good solution is found. This problem, known as "safe reinforcement learning", has so far prevented reinforcement learning from being used in safety-critical applications. Professor Nils Jansen wants to provide more safety.

In his DEUCE project (Data-Driven Verification and Learning Under Uncertainty), he wants to develop novel and data-driven verification methods that can be integrated into reinforcement learning and thus help AI to be safer. This will also increase trust in AI applications.

The Starting Grant is endowed with 1.5 million euros.


Prof. Dr. Nils Jansen heads the Chair of Artificial Intelligence and Formal Methods at Ruhr-Universität.


Bildliche beispielhafte Darstellung eines Doktorhuts
ERC Grant Projects
To top