Artificial Intelligence in the Financial Sector

AI offers several opportunities and threats for the financial sector. It already is widely used in the banking sector, e. g. chatbots, intelligent assistants for customers, automated fraud detection and face recognition software for customer identification. Furthermore, financial regulators use AI to scan Big Data for patterns and to meet their oversight obligations.

The regulation of AI in the financial sector is a balancing act. On the one hand, there is a need to ensure flexibility in order not to curb innovation and not to be left behind in international competition. On the other hand, personal rights must be protected and decision-making processes must remain comprehensible. The lack of explainability and interpretability of AI models arises primarily from intransparency in a large part of today's AI applications (black box principle).

This tension is also evident in the current regulatory approach of various authorities. On the one hand, the positive sides of AI are emphasized, such as efficiency and effectiveness gains or new methods of hazard analysis in financial market regulation. On the other hand, it is pointed out that decisions made by AI must always be the responsibility of humans and that the democratic framework must be upheld.

For the future, we see the need to further develop international regulations in a principle-based, unified and technology-neutral manner, without slowing down the development of new AI-based business models. In the global competition, Europe should take a pioneering role in regulating the use of AI and thus export its democratic values of digital freedom, self-determination and the right to information worldwide. From the perspective of companies in the financial sector, cooperation with BigTech companies could be useful in order to jointly exploit the potential of the technology in the best possible way. It would also be useful to have a common semantic metadata model to describe the data generated in the financial sector.

This publication is authored by Patrick Weber, Dr. Kevin Bauer and Prof. Dr. Oliver Hinz with the kind support of Finanz Informatik and the Competence Center for Artificial Intelligence located at Finanz Informatik Solutions Plus. We would like to thank Dr. Andreas Totok in particular for his close support during the preparation of this publication and for his suggestions regarding its content.

The Publication is available here: See the Publication