Hines kicked off the session by stating that, in financial services, it has never been technology just for the sake of technology. When it comes to AI, institutions are considering how AI is improving their operations and bottom line, as well as how to sustainably scale it. In order to get a feel for the sentiment in the room, she asked the audience which genAI use case will be the most significant in their originations over the next year.

The results of the poll were some of the most diversified across all EBAday panel sessions - showcasing the broad potential that genAI offers to financial institutions. 31% responded that streamlining operations will be the main use cases, 29% voted for improving customer service, and 25% voted for enhanced risk assessment and fraud management.

The panel discussion centred around the many use cases they are experiencing in their organisations. “Zero friction is what we aim to achieve for our clients, and the progress that we’ve managed to deliver is what we call predictive and personalised customer service.” She then described other AI use cases in their financial services wing that combine historical data with predictions when it comes to the payment journey and digital risk assessment. “We’ve managed to decrease false positives by 60%.”

Gurumurthy added that the sweet spot is a combination of AI, automation and checking processes. “It’s data extraction and classification, which helps augment the operations team. We’ve expanded it to sanctions screening and other checks as well. This has worked efficiently for the last six months, so we’re now looking to expand to client facing services.”

Schmidt agreed and highlighted the necessity to develop responsible AI. “The use case that keeps coming up is how do I, as a bank, speak to my clients in a much more personalised way? We all have the Amazon recommendations, but being able to have recommendations based on where you are in life, your savings, your mortgage rates - these are the things to think about.”

Speaking of client facing use cases, the panel highlighted the capabilities of AI in synthesising data more quickly, which is crucial in all aspects of business. Schmidt summarised: “Customer service, personalisation - that’s where AI can shine in banking.”

When it comes to challenges, Sarafidis spoke to the speed with which AI is being adopted and developed across the industry. “Chat GPT reached 100 million users in two months, that has never happened before.” He emphasised the key to scale adoption is not just leadership, and using AI to completement the human, but also the model itself. Considering 92% of AI use cases are less than a year old, he stressed that at this point of technology adoption, buying out of the box is encouraged rather than building your own solution in-house.

Lastly, the panel discussed how where AI should not be used and how heavily they expect it to be impacted by EU regulation. The panel agreed that regulation in the space is inevitable - the EU AI Act is already in the horizon - because with great capabilities come great responsibilities. Maintaining resilience and protecting data are crucial to maintain the health of the industry overall.