28 June 2021
'The digitalisation process that the bank is undergoing goes hand in hand with the use of artificial intelligence. Algorithms help us personalise services, optimise internal processes and prevent fraud and money laundering.'
Creating the future
Tim Hunter is an Artificial Intelligence (AI) Specialist within Group Innovation at ABN AMRO. He studied Artificial Intelligence and Machine Learning at the University of California, Berkeley. He started his career in Silicon Valley as a software engineer at Databricks: an open source platform on which data from various sources is brought together and processed. 'I was asked to help set up a branch in Amsterdam. That's how I came into contact with ABN AMRO and made the switch. Here, I get the opportunity to use AI to make drastic changes within the company. At Group Innovation, my colleagues and I are guiding the search for opportunities to apply AI. We are at the core of the business and are creating the bank of the future with a huge amount of drive and enthusiasm.'
Ethics
'One of my biggest challenges is to implement AI in a way that is responsible, transparent and explainable. Is a client is eligible for a loan? Are there any signs of fraudulent activity? Is this new product interesting to this client? These are all questions that AI has an answer to. But is it the right answer and may we use the data that it is based on? These are complex issues for which the bank has to find answers. Certain patterns in payment behaviour are a good sign of financial distress, but can you anticipate them? Although laws and regulations provide guidance, our guiding principle is that the use of AI may only be applied if the results are traceable and explainable and align with our own values and standards. That is why each project is assessed on an ethical level within a broad group of colleagues.'
Financial Crimes
'Many of the algorithms we use are open source. How you then train these algorithms is more important. The availability of data is crucial. For that reason, AI can be put to good use in combating fraud, which is what I am currently focusing on. Historical data teaches us to detect patterns and signs that help us spot suspicious activity. For example, we can use AI to identify groups of transactions that deviate from normal patterns. The anomalous groups of transactions that are found then need to be checked by a fraud expert. Exactly what 'anomalous' behaviour means and whether it is fraudulent is subjective and only an expert can put it into context. AI mainly leads to a boost in efficiency – it helps the expert focus on the areas that require their personal attention. And I'm proud to contribute to that.'
