While top banks roll out chatbots and virtual assistants, most mid-tier banks wait to see results before they decide to further invest in artificial intelligence (AI), according to an analyst at Gartner.
Moutusi Sau, a senior director analyst at Gartner, told Banking Dive the banking industry as a whole is still waiting to see actual value from their current AI projects before committing more resources to the emerging technology.
“Adoption-wise, banks are somewhat on a lower trend compared to the other industries,” said Sau, whose firm surveyed AI adoption strategies across several industries.
Sau said Gartner’s 2019 CIO study, which was shared with Banking Dive, found 37% of banking and investment services have deployed or plan to deploy an AI project within the next 12 months, compared to 48% in the insurance industry, 44% in retail and 40% in manufacturing.
“If you look at insurance, which is also part of financial services, they are going all in,” she said. “They are automating and doing a lot of back office work using AI, which in the banking space is still spotty. It's not picked up.”
Gartner surveyed 3,102 CIOs, including 411 from the banking and investment services industry, from April 17 to June 22 last year.
The research and advisory firm said qualified respondents were the most senior IT leaders for their overall organizations or a part of their organizations.
“The banks and investment services combined are planning lesser projects compared to other industries,” Sau said, adding another roadblock to the industry’s adoption goals is regulation.
Sau said when it comes to using AI in the industry’s big value products — mortgages, for example — regulators don't want banks using technology they themselves don't understand.
“This is not a problem that other industries face,” she said.
Sau said more communication between fintech companies and the Federal Deposit Insurance Corporation (FDIC) could foster more collaboration between tech companies and regulators, as would a push for transparency initiatives like explainable AI.
“With explainable AI, you have the exact steps in front of you, which could tell you how that machine reached that decision,” said Sau, adding, the implementation of AI technology in the loan application process, for instance, is an area where regulators often take issue.
“Regulators say that if you're rejected, there's no easy way of knowing why your loan was rejected,” she said. “But with explainable AI, that step is probably augmented, so you know why that system reached that decision.”
Another area which can be attributed to the industry’s lag in AI adoption is in its fragmentation, Sau said.
“Top tier banks are adopting AI," she said. "They're using machine learning, But that doesn’t really move the banking industry that much. What JPMorgan is doing, for example, a midsize bank would say, ‘Well, they are JPMorgan. They have the resources, the bandwidth, and the budgets to do that.’ Midsize banks are having a hard time trying to convince the CIO to allocate some budget so they can do it.”
For large banks such as Bank of America, a hefty technology budget enabled it to roll out its own AI-driven technology, Erica, last year.
Bank of America says the voice-activated virtual assistant is used by 7 million of its more than 37 million digital banking customers.
“If you think about the major types of technology that people talk about—voice recognition, artificial intelligence, machine learning, robotics—all of those apply to our industry,” said Bank of America CEO Brian Moynihan during the inaugural Brainstorm Finance conference in Montauk, New York, last month.
Moynihan, who estimated his company has spent $30 billion on code over the past eight years, added, “That’s how we reduce the size of our company, [by] applying technology across all procedures.”
But Sau says banks shouldn’t look at AI as a tool to replace jobs, but rather enhance them.
“Those kinds of jobs were already going to be replaced by technology,” she added. “For the middle office or front office, the addition of this technology is helping them target that work in a better way. … With AI and machine learning, there is improved productivity, and that improved productivity is actually helping the worker at the end of the day.”