The latest insights from your peers on the latest in Enterprise IT, straight to your inbox.
By Pragati Verma, Contributing Editor, Straight Talk
Has the uncertainty caused by the pandemic accelerated or slowed down Banks’ AI initiatives? The question — raised by Michelle Bonat, who heads AI Innovation at JP Morgan Chase for the Digital Customer Experience — was at the heart of the panel discussion about The State of AI in Banking during an Ai4 online conference that gathered data and analytics leaders from Moody's Analytics, KeyBank, Grupo Financiero Banorte, and JP Morgan Chase. She opened the discussion by asking, “How is AI faring after months of immersion into the pandemic?”
Transforming Risk Management
Asit Talukder, head of AI and ML at Moody’s Analytics responded by saying that the pandemic has made it easier for CIOs and CTOs to push their AI initiatives. “Covid-19 has changed the landscape and the risk appetite has gone up. Companies are more acceptable of trying these riskier automation solutions,” he explained.
He went on to explain why: The pandemic has changed the risk profiles of several types of customers, and banks need AI to quickly create new models that can predict a customer’s creditworthiness. “Things are moving so fast. Cash flow situation has changed, and supply chain has changed,” he said. “That’s where AI comes in and fills in the gaps because you can adopt data-driven techniques much faster.” He added that Moody’s is now “successfully using AI to analyze unstructured and alternate data to assess risk profiles much better.”
Moody’s is not alone. Mexican banking and financial services holding company, Grupo Financiero Banorte, has also stepped-up AI investments to understand the changing risk profile of their customers as the crisis evolves. “It has become an urgent matter at this juncture,” said Jose Murillo, Chief Analytics Officer. “They are good customers, and we don’t want to lose them just because they might be facing a liquidity problem now.”
Keybank, a regional bank headquartered in Cleveland, has started using AI to manage the customer experience better during the evolving crisis. According to Gaia Bellone, Head of Data Science at the bank, AI can not only respond to customers faster and better than a banker but also help a bank understand the customer better. ”Usually the data that comes from human-to-human interaction is terrible. All the interactions recorded from machines enable us to understand our customer’s data, their behavior, and their needs better,” she explained.
As they scale up their AI initiatives, almost all the panelists agreed, they are likely to attract stricter regulations. “Regulators are going to be really concerned about credit granting admission models because credit is going to be tightened in this environment and people will want to ensure that institutions don’t have any discriminatory practices,” said Grupo Financiero Banorte’s Murillo.
According to Javed Ahmed, a Senior Data Scientist at Metis, a data science and analytics training company, regulators understand a variety of technical models in areas like credit risk but might not be so familiar with the new operational risk models. “Regulation is not keeping pace with technology in several ways. Bank charters are not being granted. And new factors that fintechs are using for access to credit are not there in the equation for regulators as much,” he said. Ahmed expects threats faced by banks — from shadow banking (lending and other financial activities conducted by unregulated institutions) and fintech taking a growing share of the market, to generational changes and disruptions like Covid-19 — “will be indirectly driven by the regulatory framework.”
Murillo urged self-regulation and the idea seemed to resonate with other panelists, but they wondered what it would entail. “Faced with more regulatory attention, it will be interesting to see how we make different choices and consider options,” said JPMorgan’s Bonat. Talukder, of Moody’s Analytics, called for larger companies and medium-sized companies to form some self-regulating bodies to make sure that this is done right. “That is the fastest way to move AI adoption in finance and insurance,” he said.
The new success mantras
AI advocates in the banking world might need to adjust to much more than regulatory changes. According to Bonat, they need a new way to tell “AI stories” to senior leaders and other people outside the tech department. “Phrase it in terms of results and not in terms of technical feed you have created,” she advised.
Talukder agreed and went on add that they need to gain the trust of executive leadership and end-users. He advised two things. One is to “build trustable AI models with good data governance standards” that are “rock-solid in terms of privacy and security.” The second is to align AI projects to business goals. “Prioritize your AI initiatives in terms of whether you are looking for revenue growth or improved efficiency,” he said.
Ahmed, of Metis, began his argument by pointing out the problems that AI projects typically face. “There is a bias towards things with more bells and whistles and more math behind them. People think a more complicated approach is going to deliver better results. That’s not always the case,” he asserted. To him, it is important to understand when a certain approach or model will work and when it won’t. Models used in a typical forecasting exercise, he said, “rely on the assumption that the past and the future are governed by the same set of principles, environment, and forces.” He expects more such models to fail because of the massive shifts in the post-pandemic world. “In this new world of Covid, there will be more failures of models and more calls for models that incorporate judgment to account for factors that we haven’t yet seen and don’t know how to quantify. It’s important that decision-makers understand how these models work and their limitations,” he said.
Future of Banking
Looking forward, all panelists predicted a rapid advancement of AI technologies in the financial world. Ahmed and KeyBank’s Bellone expect fintech companies will take the lead. “AI innovation will be more and more concentrated in fintechs….And big organizations like ours [will] integrate their solutions in our processes,” said Bellone.
Murillo and Talukder, however, asserted that big institutions will adopt AI very aggressively. “Large institutions have a lot of initiatives on the way,” said Talukder. Within the next five years, he expects AI to create compelling customer experiences and hyper-personalization at the front end, while enabling “a more comprehensive view of risk across the institution, right from KYC compliance to supply chain risks. “What I see is the impact of AI across the board from the front office to the middle and back office,” he said.
If there was one clear message from this panel, it was that whether fintech or big banks drive it, AI is all set to shape the future of the financial world.
Pragati Verma is a writer and editor exploring new and emerging technologies. She has been a business journalist and managed technology sections at India’s The Economic Times and The Financial Express.