AI adoption in the Luxembourg financial sector

The report's findings are based on a survey conducted by the CSSF in collaboration with the Luxembourg Central Bank between October 2021 and January 2022.

Some of the key findings of the report are:

• The level of adoption of AI and other innovative technologies in the Luxembourg financial sector was fairly limited and still at an early stage at the time of the survey, with only 32% of respondents investing in AI and only 14% of respondents investing in distributed ledger technologies.
• The most prevalent use cases of AI were related to customer relationship management, risk management and compliance.
• The main anticipated benefits of using AI were improved internal process efficiency, enhanced fraud detection and improved products and services offered.
• The main challenges faced by the institutions using AI were data quality and availability, regulatory uncertainty, lack of skills and resources and ethical concerns.
• The majority of deployed AI solutions had some form of human oversight or intervention in the AI decision-making process.

A total of 158 different use cases for AI technologies were reported, with 59% of them being in production. The top five areas of use cases were AML/Fraud detection (18%), Process automation (15%), Marketing/Product recommendation (8%), Customer insights (8%), and Cybersecurity (8%). To a lesser extent, AI technology is also being used for sentiment analysis, credit scoring, customer support, algorithmic trading, and robo-advisors. Development in some of these areas is more advanced than others, with AI process automation and cybersecurity solutions already being deployed, while development of AI robo-advisors and IRB credit risk modelling solutions is still in its infancy. This can be explained by the fact that in the areas of high adoption such as cybersecurity and AML/Fraud detection, respondents rely mainly on third-party off-the-shelf solutions.

Regarding development approach, 73% of reported use case solutions were developed in-house. However, the report found that a large majority of respondents do not have any specific AI-related governance mechanisms such as an AI ethics policy or an AI ethics committee. The Report highlighting that oversight of the AI development process is at the moment mostly handled by the Data protection function, the IT Security function, and to a lesser extent by the Risk function. At the same time, 77% of all reported use cases have a “human in the loop” and 51% implemented bias prevention/detection techniques. A large majority of respondents indicating that their underlying AI models have good auditability and explainability.

The survey also shows that there has been a general increase in investments across all categories of innovative technologies in the 2022-2023 budget compared to the 2021 budget, with the most significant increase in the area of machine learning.

The survey demonstrated that although AI usage in the Luxembourg financial sector is currently limited and in its early stages, investments in this technology, particularly in machine learning, are expected to rise, paving the way for more widespread adoption of these innovative technologies in the near future.

The full CSSF Thematic review on the use of Artificial Intelligence in the Luxembourg Financial sector can be found here.