Financial services providers are failing to successfully implement artificial intelligence, claim European fintech executives, despite growing evidence that the much-vaunted technology will increase productivity and reduce costs..
Fear of job loss, regulatory concerns and institutional inertia are among the many aspects stopping bankers from fully leveraging the systems underlying products like ChatGPT.
“The big banks will certainly not adopt (the technology) as quickly as the opposite fintechs,” says Tom Blomfield, co-founder of Monzo and partner at Silicon Valley start-up incubator Y Combinator. However, generative AI will make banks “more efficient and enable them to supply the identical products at a lower cost.”
According to a study by Capgemini, only 6 percent of retail banks are able to implement AI on a big scale of their business. However, McKinsey estimates that this might add as much as $340 billion in value to the worldwide banking sector annually, which is about 4.7 percent of the industry's total revenue.
Many imagine the technology, which might answer questions and analyze huge amounts of text and numbers in seconds, can dramatically reduce costs across the industry. But there are fears the disruption could lead on to job losses.
“People don't understand that it serves as a productivity tool,” says Nasir Zubairi, managing director of the fintech accelerator Luxembourg House of Financial Technology. “They still imagine it would take their jobs away.”
He added: “Traditional banks are fundamentally analogue and the transition from analogue to digital has all the time been a difficult thing.”
Speaking on the Financial Times' TNW technology conference this month, Zubairi cited anti-money laundering audits for instance, where institutions typically employ staff to comb through spreadsheets searching for unusual activity.
He said that when he demonstrated to an establishment how this may very well be improved with a customized AI model and that he estimated that this may “immediately save as much as 450,000 euros in salaries per 12 months,” it was rejected.
“People don't like firing people,” he added. “They wish to protect the function of their job, and in the event that they have to fireplace people from their team who’re doing those jobs, they may additionally be threatened as management or their power could also be undermined indirectly.”
According to the Bank for International Settlements, central banks have recently been encouraged to step up their efforts in the sphere of AI, saying that while the technology can result in productivity gains, it also carries risks, corresponding to the sharing of false information and vulnerability to hacking.
A typical problem with large language models, the technology behind most generative AI products, is their tendency to “hallucinate,” or present inaccuracies as fact. They are also known to generate information based on the info they were trained with, raising concerns about sensitive or secure information.
“There isn’t necessarily opposition to (AI), but there are concerns,” said Wincie Wong, digital chief at NatWest, who called for the technology's risks, ethics and vulnerabilities to be assessed before deployment. “Ultimately, we’re considered one of the massive banks and many purchasers keep their data and their funds secure with us. We need to respect that.”
Customer service is considered one of the areas most disrupted by AI tools that may communicate and reply to queries in a human-like manner. For greater than a decade, digital banks have used machine learning to prioritize online queries and sometimes route customers to a live customer support agent.
However, LLM-powered bots can understand a wider range of requests no matter how they’re worded and execute decisions, corresponding to ordering a bank card, eliminating the necessity for human intervention.
“I actually imagine that this may lead to the lack of the overwhelming majority of customer support jobs over the subsequent 12 months to 5 years,” said Monzo's Blomfield.
Many banks and fintechs, including Klarna and NatWest, are already using AI chatbots for customer support. NatWest's Wong said they’d made great progress with generative AI of their service AI Cora, receiving greater than 11 million chats in the course of the 12 months, with greater than half requiring no human intervention. In 2017, the service was receiving 1,000 chats monthly and requiring intervention.
Swedish fintech company Klarna said its AI assistant can do the work of 700 customer support agents and resolve queries in under two minutes, in comparison with 11 minutes previously. As a result, the corporate expects to avoid wasting $40 million in customer support costs this 12 months.
But Wong said that training the models to acknowledge nuances is critical to success, corresponding to understanding that an address change can have an emotional undertone, corresponding to a bereavement within the family.
“It was really vital to know the psychology behind it. And when you don't do it right, quite frankly, you may upset plenty of customers,” she added.
In addition, when introducing the brand new technology, banks needed to be careful to stick to the industry's strict compliance rules and navigate an unfamiliar regulatory environment.
In a landmark 2022 ruling, a Dutch court ruled in favor of neobank Bunq after it sued the Dutch central bank for banning it from using AI to conduct money laundering checks.
Regulators lifted restrictions on German fintech firm N26 last month after it tightened its controls. For years, the bank had capped the number of latest customers it could tackle due to its poor anti-money laundering controls and needed to pay hundreds of thousands in fines for repeatedly being late in submitting reports on suspicious activity.
Carina Kozole, chief risk officer at N26, said the corporate has worked closely with regulators to develop an AI model that may determine whether a brand new customer is a criminal, which has led to a 90 percent reduction in cases on the platform.
“If we don't adopt AI within the industry, we won't be here in just a few years,” she added. “We have to reveal the advantages and the way we are able to change into compliant after we use AI.”
