The views and opinions expressed in this article are those of the thought leader as an individual, and are not attributed to CeFPro or any particular organization.
Michiel van Lunsen, Data Science Model Validator, ING Bank
The data science model validation chapter focusses on the domains: Loan Pricing, Analytics, KYC Fraud & Compliance. The majority of these models use machine learning techniques.
AI and ML models are more sophisticated in finding patterns in data compared to traditional models. This is especially important in the KYC and Fraud domain. Criminals are becoming more advanced in their approach, so we need more sophisticated models that capture rapidly changing behaviour.
AI and ML are rapidly evolving fields. That means that we need to keep up with the latest technologies, but also be aware of any limitations that these new technologies have. Furthermore, there is a challenge when there is a combination of expert rules and machine learning and how to approach “black box” vendor solutions. We’re currently in the process of establishing the guidance on this specifically.
In ING we want to stay one step ahead. AI and ML offer a great improvement to more traditional models. However, that added complexity can lead to bias. It is our responsibility to make sure models within ING are used ethically.
We now see more and more awareness of ethics in banks. I expect that in the future consumers will also make their decisions more on the ethical and environmental
Michiel van Lunsen will be presenting at the Fraud & Financial Crime Summit, this event will be taking place on September 20-21 at One America Square.