AI in banking – game changer or game challenger?
Euromoney, is part of the Delinian Group, Delinian Limited, 4 Bouverie Street, London, EC4Y 8AX, Registered in England & Wales, Company number 00954730
Copyright © Delinian Limited and its affiliated companies 2024
Accessibility | Terms of Use | Privacy Policy | Modern Slavery Statement
Sponsored Content

AI in banking – game changer or game challenger?

Sponsored by

new-caixabank logo.png

Artificial Intelligence (AI) promises to transform financial services, enabling banks to tailor products and services to customers’ needs at the times they need them; but success means solving both technology issues and challenges around security, privacy and ethics.


700x400caixadec19

Today, retail banking customers can choose from a range of personalized products and services. However, it is critical that customers understand the choices they make and are offered those that are most appropriate.

From data to insights

To do this, banks must be able to collect and analyze enough information on their customers’ financial situation. The first part, collection, is not usually a problem. Via their traditional, and newer digital, channels, banks record almost the entire financial lives of their customers, from wages to purchase preferences. But analyzing this abundance of raw data to generate insights accurate enough to make appropriate financial recommendations is a different matter.

At CaixaBank, a European leader in digital banking, AI solutions are already being applied to every part of the customer journey, from onboarding and product advice to suggestions on credit card payments. For example, CaixaBank’s new omnichannel relationship model, inTouch, uses sophisticated data analytics to tailor each step of the customer journey to the individual user. First, those algorithms help to target potential customers for onboarding, resulting in a 93% acceptance rate. Then, continued data modelling is used to target personalized products and services appropriately, raising the number of products purchased by each customer.

Beyond background algorithms

AI can also deliver advice and suggestions directly, not simply via algorithms working in the background. CaixaBank’s various chatbots can, for example, answer up to 1,000 different voice or text questions and proactively suggest personalized products and services based on individual profiles and behaviour. They can even suggest the best ways for customers to make their credit payments.

Extending this idea, CaixaBank’s Smart Money ‘robo-advisor’ is a 100% online solution that delivers personalized investment recommendations and management. In just 18 months, it has attracted 44,000 portfolios with a balance of €971 million. As AI develops, these robo-advisors will get smarter at monitoring external trends and analyzing their own clients’ financial habits, delivering even more personalized recommendations.

Most recently, CaixaBank has used AI to manage returned direct debit payments. The bank is the primary direct debit service provider in Spain, managing 450 million direct debit payments every year, and a key issue is how to handle debits when an account has insufficient balance. Previously, a human advisor would decide whether to allow, reject or postpone the debit. Now, the bank is using an AI system to make these decisions and resolve incidents with 99% accuracy.

Getting AI right

Clearly AI has the potential to deliver huge benefits for both customers and their banks. But, as always, these benefits come with potential risks. 

Firstly, AI solutions must work. That is, if an algorithm is recommending financial products, then it must recommend them appropriately. If it goes wrong, it will go wrong for a lot of people quickly. So, the client data on which the algorithms rely must be correct and up-to-date. And the algorithms themselves must “understand” suitability and, if they are true AI, they must be parameterized to prevent them evolving beyond their original remit. 

More subtly, AI solutions need to “work” ethically. The first ethics problem is that without huge amounts of data to train the models, there can be no accurate AI; but if that data is biased, the models’ conclusions will repeat the biases embedded in the data. So, the datasets used to train self-learning models must be adapted so as not to perpetuate the undesirable biases they may record.

Pere Nebot, chief information officer at CaixaBank, acknowledges these issues: “There are biases in society and we ensure that none of them are translated into our algorithms to avoid reinforcing these biases through automatic decisions. So, we are investing in this area and participating in industry initiatives around an ethical framework for the use of AI.”

The next ethical issue concerns data privacy in its broadest sense. This means not just unauthorized access to data, but unauthorized use of data to promote products and services. Banks must therefore be sure that the aggregation and processing of client data to produce product recommendations is sanctioned by customers and that they are able to opt out. 

Predicting the future

These issues will be joined by others as AI solutions become smarter. Victor Allende, executive director of CaixaBank Private Banking, says: “In the near future, wealth managers will have access to a large amount of data that can be used to design personal investment journeys not only for each individual client, but also their children or other family members, depending on their specific habits, financial situations and personal lives.”

Ensuring the best customer outcomes while preserving privacy and security in such a data-driven environment will be one of banking’s biggest challenges. 

Gift this article