A creative programme that supports software engineers in their work, an algorithm that formulates helpful reports for bank customers from a jumble of market data, a tool that assesses risks and thinks outside the box – the list of possible applications for artificial intelligence (AI) is "manifold", explains Bernd Leukert, Member of the Management Board responsible for Technology, Data and Innovation at Deutsche Bank. However, the bank is testing artificial intelligence for specific applications. There is no "big bang", i.e. an abrupt and broad introduction, he said on Thursday in an evening discussion organised by the Association of German Banks in Frankfurt's Techquartier.

Rupert Schaefer, Executive Director for Strategy, Policy and Governance at the German financial supervisory authority BaFin, also sees AI at an early stage. Processes for analysing large amounts of data (big data) are already in use, as are learning algorithms (machine learning), for example, in operational tasks such as risk management, high-frequency trading or portfolio management. The situation is different for so-called generative AI, which can hold conversations or formulate independent texts and create content, i.e. is creative. "We don't have the impression that much progress has been made in terms of implementation."

If the term artificial intelligence is defined more broadly, however, progress is already clearly visible. Around two-thirds of the banks, insurers, and other financial service providers surveyed are already using various processes in practice, emphasised Andreas Hufenstuhl, Partner for Big Data and Advanced Analytics at the auditing and consulting firm PwC, in the panel discussion.

He referred to a survey conducted by the consultants: Social media channel management, sales analyses, fraud management, money laundering checks, and in-depth data analyses are among the most common practical applications. "This is no longer in experimental status but has already arrived in day-to-day operations," said Hufenstuhl about the various applications.

The EU Commission is currently preparing a framework for artificial intelligence. For example, it introduces a right of complaint and information for consumers, broadly bans controversial fields of application such as facial recognition and creates a new category for generative AI. The aim is not to regulate individual use cases in detail but to introduce principles for dealing with AI, said Jan Ceyssens, who is responsible for digital finance at the EU Commission. The "AI Act" is expected to be passed before the European elections in June.

A standardised legal framework is essential, praised Deutsche Bank board member Leukert. "In this respect, we played a pioneering role in Europe." Now, it is essential to create standardised international rules, which requires political willingness. Deutsche Bank is relying on partnerships with Google and Nvidia and is optimistic about the new technology. Not only the bank but society as a whole will benefit from it.

However, there are also significant concerns: Companies need to adapt to faster software development cycles, warned PwC partner Hufenstuhl, as programming is also much easier with the help of artificial intelligence. At the same time, companies need to take the workforce with them and not place too much in the hands of AI. "Especially in the context of generative AI, it is vital that we don't take people out of the loop."

Both inertia and excessive use of AI can be a problem, as the discussion showed. BaFin Executive Director Schaefer smugly remarked on supervisory processes that are characterised by "legacy". "We still have paper." However, Leukert warned that AI should not be used for everything out of convenience. The use of computing power costs money. "The use of AI is also a temptation."