This audio is automatically generated. Please let us know if you have any feedback.
The banking industry quickly recognized the business potential of generating AI, and on the other side highly valued the risks inherent in reckless recruitment. The largest institutions in the sector are proficient in risk management. This took a cautious yet sustainable approach to moving pilots into production.
Recruitment has gained momentum last year, according to clear insights tracking 50 of the biggest banks in North America, Europe and Asia. The 50 banks announced 266 AI use cases last week, up from 167 in February, Intelligence's Colin Gilbert said Tuesday in a virtual roundtable hosted by industry analyst firm.
“The majority, or about 75%, are still facing internal or employees,” he said, adding that the distribution between the generator AI and traditional predictive AI use cases was split by about 50/50.
As banks integrate technology into their daily operations and mature their models, the mix is shifting towards generic AI capabilities with capabilities aimed at customers. MuditGupta, Partner and Americas Financial Services Consulting Practice AI Lead said during the panel.
“Because of the low risk, we tend to start with productivity,” Gupta said. “As we go further down the adoption path, we establish proof points so that we can proceed to change.”
The three global banks' technology executives each put their own spin on the development of Gupta.
“We're taking progressive steps to do exponential things,” said Rohit Dhawan, director of AI and director of Advanced Analytics at Lloyds Banking Group. The bank is using Oracle's Azure-based database system and Exadata Customer Cloud Data System to integrate AI efforts to move beyond individual use cases after strengthening its cloud-based data strategy earlier this year.
“This is a very different idea. It's a very different idea, from thinking about how to inject or optimize processes with AI, to radically rethink processes with AI,” says Dhawan.
There are many generative AI use cases in banking. The technology has the capabilities spanning the entire process, from managing the vast amount of customer and compliance data of associates to help engineers refactor legacy applications.
Bank executives expect generative AI to handle up to 40% of daily tasks by the end of the year, according to an April KPMG report. Three in five of the 200 US bank executives surveyed by the company said the technology is essential for long-term innovation plans.
AI Acceleration
Until recently, Natwest Group was slowly moving along with AI to measure return on investment by one use case at a time, Zachery Anderson, the bank's chief data and analytics officer, said during the panel. “We have begun to rethink the works that have really seen the customer experience in the past eight months, especially those that have started to rethink how to completely reconstruct them back and forth,” he said.
AI assistants, such as helping Bank of America employees and City's stylus document intelligence and virtual assistants, are becoming more common, but as deployments increase, the range of technology capabilities is expanding. In September, JPMorgan Chase announced that it would equip 140,000 employees with LLM Suite AI Assistants.
“Generative AI affects every part of the bank, every part of the job,” Accenture Global Banking Lead and Senior Managing Director Michael Abbott told CIO Dive in January.
Natwest is taking advantage of two deployment pathways, Anderson said. “We have a core set of data scientists, data engineers working on the biggest and most challenging use cases,” he said. “They are currently working on what is currently on the edge of feasibility as the model is improving so quickly, but by the time the project is finished, what was previously at the core of the possibilities now.”
At the same time, banks are pushing AI into non-technical functions. In addition to providing tools to developers, NatWest has deployed internal AI tools for business users and “a very large portion of bank users.”
“The viable edge of what models and agents can do is increase, as well as the jagged ones,” Anderson said. “You end up realizing that you think you can do, that you can't, and that you can sometimes surprise you… As all of our employees are exploring that edge, we're mapping frontiers in a much faster way than before.”
Truist has moved from a quick win to a use case that further reaches the bank's food chain. “Knowledge extraction is the most popular use case,” said Chandra Kapireddy, Head of Analytics for AI/ML and GEN AI at Truist. “It's a really low risk. The data is already there and it's a high reward [because] You get the answer quite quickly. ”
Answers will help communicate value to business users and help maintain momentum as AI use cases increase in complexity and cost. Also, early victory provides political capital to IT executives to engage in technically necessary experiments.
“If you try to be perfect, you're turning the wheel,” says Kapireddy. “It will be very productive at the start of the use case lifecycle. But once you start investing the costs in it, you need to make sure that the stakeholders in the business know that it will have an impact.”
