Bart Leurs (Rabobank): ‘We Implemented a Temporary AI Stop’

You are Chief Innovation and Technology Officer. Many organizations have a Chief Information Officer or a Chief Technology Officer. Why is that role combined at Rabobank?
‘In our vision, technology and data are so strongly linked, you cannot divide it into two separate roles. Business processes nowadays mainly run on data and technology; there is less and less manual intervention. Innovation does not yet always form part of it. That is the case at Rabobank: our thinking is that many innovations are driven by technology and data.’

How does the agile organizational structure contribute to the innovations within Rabobank?
‘Some 7,000 people now work in tribes and squads, as small businesses they contribute to the customer’s digital experience. Decision-making lies lower. Because they operate close to the business, they have a better understanding of what is going on and how best to innovate. Developers and data scientists work closely with the tribes to further improve the banking app, for example. Technology and data thus run like a thread through the organization.’

What opportunities do you think AI technology offers?
‘Those are huge. I am very excited about how AI can contribute to solving major social issues. There are also endless opportunities within our organization. But everything hinges on solid data management and data governance. 
There is no domain where AI technology does not impact the bank - whether it is sustainability or detecting fraudulent practices. The foundation must be right. Only then can we compete at Champions League level with major innovations. Data governance is extremely important in this regard. We must ensure that our 42,000 people handle data in the right way and that ultimately there is reliable and high-quality data on our platform.’

How do you ensure good data governance?
‘We invest in it heavily and launched a master plan for data management, for example. It means, first, that we have all employees take data analytics training courses. Colleagues need to understand the importance of data and the role it plays. Data and technology are also a strategic learning point within the Executive Board because both areas of expertise are crucial to running our company. We cannot manage them from a corner of the organization. An important additional step is for us to invest in technology that is future proof. On our new Global Data Platform, we will soon have all the company’s data in one place and will be able to load limitless data.’

What are the biggest challenges when it comes to data?
‘For data to be used optimally, it must be accessible and structured. We need to ensure good and secure technology, and, in addition, data must be of good quality and be reliable. Ultimately, with the help of algorithms, we make decisions based on this data. You approve a mortgage or not, you classify someone as a fraudster or not.
I think we have come a long way. Many of our core processes already run on data and algorithms. Fewer and fewer decisions are made by humans. At the same time, the models must be good; there must be no bias in them. Also, explainable AI is important. If a customer is denied a mortgage, we must always be able to explain why that is so.’

How do you ensure that data are qualitatively in order?
‘For the past two years we have been working intensively on data governance. We are still not done with it. In the past, fields in our systems have been filled in incorrectly - think of a note from a mortgage broker or the use of a comma instead of a period. There was no realization yet that someday this data would be used for algorithms. A good number of employees have been working to fix those data fields. It is painstaking work, but we cannot escape getting the data at the desired level of quality.’

Rabobank’s mission is to grow a better world together. How does your organization innovate to deliver on that promise?
‘I was recruited six years ago to drive innovation within the bank. That was quite a quest in the beginning. Initially, we started very broadly. For example, we introduced SurePay: for online payments, the IBAN number is automatically checked against the name of the account holder. And Peaks: an investment app that allows consumers to invest with their small change. We did improve our services with these innovative products, but we did not really distinguish ourselves.
That led to a period of reflection - what makes us special and what innovations do we want to focus on? Gradually we discovered: to be able to innovate with impact, we need to collaborate with partners. It fitted well with the cooperative idea. We also chose to, going forward, link innovations to three transitions that we want to accelerate based on our strategy. The first is the food and agriculture transition: how do we produce enough food in a sustainable, responsible manner for the growing world population. Secondly, we are committed to the energy transition and an inclusive society. We have now realized 42 innovations in these areas, most of them in food & agriculture. That makes us as a retail bank different from ING or ABN AMRO.’

Can you give an example of an innovation in the food & agriculture sector?
‘There are several great projects in the pipeline, but one of our best projects so far is Acorn. With this we encourage small farmers in South America, Asia, and Africa to grow trees on their land. They often struggle to generate sufficient income. With the Acorn project, they earn extra by growing mangoes or cashew nuts, for example, in addition to corn. Because these trees remove CO2 from the air, these farmers receive CO2 rights that they can trade to multinationals on an international platform. For now, 60,000 farmers have joined. That number is growing, in part because development organizations, such as Solidaridad, for example, are assisting farmers to join the platform.
Acorn is a collaboration between Rabobank and Microsoft. We have access to thousands of farmers; Microsoft built the platform. By joining forces, we are doing something good for the world. Not only do we reduce overall CO2 emissions, but we also help farmers gain additional income. The project illustrates how data and technology can move the world forward. We employ more than 40 satellite specialists who can accurately estimate whether trees can grow well in certain areas and what the future expectations are.
It is quite remarkable: in the past, when granting a loan, we relied on financial information from the farmer. Now we have much more relevant data, which we will also increasingly need to achieve our ambitions. We are already far advanced in satellite technology, in, for example, analyzing land worldwide. We thus created a digital twin - a virtual representation - of the earth. Thanks to real-time data, we thus monitor the health of our planet.’

Which data and algorithms will become decisive in decision-making in the future?
‘In general, we look beyond mere financial data. We are a gatekeeper of the financial system; we have a responsibility to spot fraudulent practices. Thanks to AI, we will become even better at that. In addition, we will increasingly look at a company’s sustainability performance or diversity policy when approving a credit application.
In the field of food & agriculture, data sources are becoming ever more expansive, for example on climate development. What we are also focusing on increasingly is traceability. Consumers want to know where the tilapia fillet on the shelf comes from, or where the mango was picked. This is becoming more important to us too. I think there will be many more requirements from governments regarding traceability.’

We may only be at the beginning of an evolution where data can help make the world a better place. What is your expectation?
‘The importance of data and algorithms will grow exponentially. This is partly due to the lightning-fast developments in GenAI, or generative AI; complex technology has become so much more accessible. Thanks to prompt engineering, you no longer need to be able to code to develop software; you need only be able to issue commands. GenAI also helps developers. With a program like GitHub Copilot, they can code better, faster and more error-free. While writing code, the developer gets suggestions for new lines of code. The program even generates complete functions. That acceleration is going fast, so fast that we sometimes cannot control it.’

That is also often the societal fear - that we can no longer control AI.
‘I share that concern. I think governments and companies have a duty to ensure that we deal with AI in a good, ethical way. From our Executive Board, we have therefore recently declared a temporary AI ban. We saw that developers started experimenting very enthusiastically with, for example, ChatGPT and Google BARD, also an AI tool to generate codes. That is of course great as it will lead to interesting innovations. But, as a bank, we do not want certain codes to be shared or customer data to end up on platforms of American tech companies. Neither should we, our customer data is well protected. That is why no one is currently allowed to interact with ChatGPT or Google BARD from corporate devices. We want to quickly identify what interventions are needed to mitigate the risks. How we can use these tools in a secure environment. Then we can swiftly take up the opportunities that AI provides.
I expect every company to think carefully about the risks of AI and take appropriate measures. In addition, governments must create a good regulatory framework. Because things can go horribly wrong with AI. I understand that the future AI Act is well designed. This European Commission law to prevent undesirable AI practices is still being fine-tuned here and there.’ Jokingly, ‘Perhaps with ChatGPT’s help.’

Managing data and using AI takes a huge amount of energy. How does Rabobank deal with that?
‘We recently introduced sustainable coding. We want every engineer to understand that CO2 is created when he or she is coding. That awareness can be improved. That is why we have made someone within each team responsible for energy consumption. We are indeed going to measure that. In part we can do it ourselves, and in part we depend on data centers. That information is still not shared enough. I assume that companies will request more of this information and include it in their sustainable finance reports. There is no other way.’

Interview by Koko Visser, partner at Valcon. Published in Management Scope 06 2023.