The banking sector is currently dealing with Anti-Money Laundering (AML) issues, with stricter legislation in Know Your Customer (AML) being the instigator. De Nederlandsche Bank is pursuing fully explainable and transparent models with a risk-based approach. Councyl provides such a model: Behavioural Artificial Intelligence Technology (BAIT). This study aims to analyse the current environment BAIT would be implemented, to see if the tool fits the requirements of the banking sector. The inherent features of BAIT, adding to the four functions make it an interesting tool for banks to consider.

For decades researchers deliberated and continue to debate on how to support and assist humans in decision-making. This resulted in the development of Intelligent Decisions Support Systems (IDSSs). An IDSS is an application of artificial intelligence (AI) that desires to enhance and support decision making by enabling tasks to be performed by a computer while mimicking human capabilities. The two most generally classified types of IDSSs are knowledge-based and non-knowledge based systems. A knowledge-based system, also called an Expert system, directly translates domain knowledge into a set of rules or cases to support human decisions. In contrast, non-knowledge based IDSSs apply the rapidly growing branch of AI known as machine learning (ML), that grounds its decision-support on feature extraction of labelled training data. Recently, a company called Councyl, in collaboration with the TU Delft, developed a new approach to AI that has the potential to constitute a novel type of IDSS for judgement purposes. The new approach to AI is called BAIT (Behavioural artificial intelligence technology). BAIT utilises discrete choice modelling (DCM) to codify the domain expertise of experts’ in order to provide introspection on their decisions and support future judgments.
As BAIT is a new IDDS approach it requires testing in different settings to gain insight into the usefulness and effectiveness of this new method.