Decoding AI in financial services
A new book out by Clara Durodié is a must-read for financial institutions grappling with both the challenges and opportunities that AI has brought to the industry. Efma’s Boris Plantier spoke with Ms. Durodié about the book and AI’s impact in financial services.
What compelled you to write this book?
Artificial Intelligence (AI) is rapidly changing our world - from the shape of our society to our customers’ habits. Covid-19 has accelerated this trend at an unprecedented pace. Fintech is advancing faster into banks’ traditional market share with solutions based on AI systems. But the vast majority of leaders at banks have very little comfort in making dramatic decisions about AI. Historically, they have never had to get too involved with the technology, but now it is essential to have an AI strategy. The more informed they are, the stronger their mandate for change is going to be.
Therefore, it is imperative that banking leaders have access to an impartial reference guide which is easy to follow, with practical questions about what artificial intelligence is, and what it means to business, governance, and profitability models.
AI is a complex field, marred with risks and numerous challenges when deployed in real life. Therefore, it is ever more important for Boards to educate themselves about the risks so they can manage adequately. The book addresses the AI educational gap at the board level. The book empowers decision makers with a strong AI foundational knowledge to ask confident and informed questions about the investments and use of AI.
There’s a lot of fears concerning AI and data. People seems to fear a 1984 scenario. Are these concerns well-founded? Or are they overblown?
The risks are real. In our industry, they are multiplied 100x. It boils down to one simple point: ask the hard questions to inform the right decisions. Making costly mistakes can carry a high reputational risk and business damage.
There are many futurists who are very excited about AI. They have been hired to advise regulators and decisions makers. But, these futurists are typically not cognizant of the highly regulated environment in which we operate. Therefore, their advice is not always suitable. “Break things now, apologize later” might work in Silicon Valley but our industry operates very differently. We need a well-considered approach of every AI system we procure into our businesses. This approach needs to address the balance between innovation and regulatory requirements.
Our industry operates with some of the most valuable data: financial data (private and institutional). This data can be a superpower. Therefore, it is our job to learn its value and protect it.
What are the key hurdles to AI adoption in banking?
Education on AI is the key hurdle. Why? When people have a foundational knowledge of what it is, then their mindset changes. I have seen how the organizational culture improves. People also get better at using AI, at identifying AI errors, and their vocabulary changes - it is well informed, knowledgeable, and they can sustain a sound discussion. The conversation becomes informed and fluent.
We need thorough and practical education on adoption of AI. But not the branded online courses from leading academics. That is not real life. That is the business world seen through the lenses of a scholar not of a practitioner. Therefore, those courses don’t add any real value.
We need to educate all of our staff — from the receptionist to the board director - about what AI means to our business. When we involve everyone, the AI transformation comes easier. One board director told me: “change is about winning hearts and minds. How do we do that with AI?” And the answer is: through practical information and education.
How can banks use AI and data to help their customers in their banking journey?
For the first time in our industry’s history, we are able to provide personalization at scale for our clients. This is the greatest advantage of AI. The biggest risk is that we don’t have a suitable regulatory framework to regulate the personal financial data and financial data in general. The regulators are working very hard to put the systems in place, and I have worked with teams who are inspiring and visionary.
So, how to do it? Firstly, banks need to satisfy the preconditions of bringing AI into the enterprise. This is the AI infrastructure, which I wrote about at great length in my book. They also need to articulate their data strategy, an essential piece which also addresses the myriad challenges in data management. Then, they need to progress with the selection and adoption of the most suitable AI systems that support the business objectives and satisfy regulatory requirements including privacy considerations. At each stage, one needs to have people who really understand what is being built and the associated risks. Institutions needs to have people in management positions who are informed and up-to-date on recent advancements in AI. For instance, since 2018 we registered great developments in transfer learning (an approach to machine learning) which addresses privacy concerns.
We often hear about data monetization or cost reduction as potential benefits of AI.
I have never been an advocate of data monetization. There is a booming industry on data brokerage and monetization, but I believe in financial services we need to be extra careful. We are in uncharted waters.
AI brings efficiencies and as a result, companies can attain cost reductions. But, let’s not allow ourselves to be lured by short-termism. Moreover, there is a limit as to how much we can reduce costs. Therefore, I have always advised that our industry’s leaders should regard AI as a strategic business tool rather than a mere cost reduction tool. When AI is correctly selected and built, it enables companies to create new business models, new revenue models, and find novel ways to compete and attract clients.
What are you expecting from AI in the near future?
AI is a transformational force in our industry. We have seen the early adopters who rushed to build the right infrastructure to enable them to capitalize on their investments and gain a competitive edge.
We are now seeing a new wave of AI adoption that of ethical and trusted AI. As companies move deeper into an ESG agenda, and governance and purpose shape the narrative, ethical and trusted AI is the tool to deliver on these objectives. Ethical and trusted AI is not an empty narrative. We have demonstrated that ethical and trusted AI increases profitability while reducing costs, and it is a future-proof strategy to invest in technology which will carry us for the next decade and beyond.
Decoding AI in Financial Services is available now! Better yet, Efma readers receive an exclusive 25% off when they order – just enter EFMA25 in the checkout to receive your discount!