In June, Citigroup published a research report that predicted artificial intelligence will displace 54% of jobs in the banking industry, more than in any other sector. A Bloomberg Intelligence report released Thursday found that global banks are expected to cut as many as 200,000 jobs in the next three to five years as AI takes on more tasks.
Experts and anecdotal evidence so far suggest generative AI is changing jobs in banks, but not killing them.
"I'm highly confident it's going to change the way we work, but I think it's going to create different types of work," said Mike Abbott, global banking lead at Accenture, in an interview. He estimates AI will affect 75% of banking jobs in some way.
The advent of generative AI is kind of like the impact Microsoft Excel had when it came out in 1980, Abbott said.
"Everyone said it's going to eliminate the finance people," Abbott said. "I think it's pretty safe to say we have not eliminated very many finance people since 1980, despite the introduction of Excel. Because what happens is it unleashes creativity around things that you couldn't do before. Generative AI will open up a whole new class of opportunities and job descriptions and capabilities that we can only imagine today."
It's already begun creating new jobs and ways of working for people throughout financial services.
New AI jobs being created
Banks have been both promoting existing executives to positions that include AI oversight and hiring for entirely new AI jobs.
Big-bank executives with AI oversight include Chintan Mehta, CIO for digital, innovation and strategy at Wells Fargo; Dominic Venturo, chief digital officer at U.S. Bank; and Teresa Heitsenrether, chief data and analytics officer at JPMorgan Chase. "We're seeing the emergence of chief data and analytics officers," Abbott said.
"Titles matter in banks," said Daniel Latimore, chief research officer at The Financial Revolutionist. "Reminiscent of the innovation craze several years ago, the surge in AI-related positions shows the emphasis banks are placing on this transformative technology. Over time, AI, like innovation, may become everyone's job."
One relatively new AI job title in banking is "prompt engineer" — a person who creates text-based prompts or cues that can be interpreted and understood by large language models and generative AI tools. Morgan Stanley has at least two, for instance. This position is also sometimes called "AI whisperer."
"We're seeing prompt engineers, especially in the operations side," Abbott said. In technology departments he sees more "product engineers" — "somebody who is a product owner and a prompt engineer at the same time, kind of the tech and business merging together."
Other new jobs generative AI adoption is opening up in banking include model tuners and model trainers who program settings of AI models and manage the training data that they receive (often through a technique called retrieval augmented generation), model validators, model risk managers and de-biasers (people who analyze models being developed and make sure that they don't have any biases in them).
Some tech companies, including Microsoft and General Assembly, have AI ethicists who are supposed to ensure that the company's AI technologies are developed and deployed in an ethical manner. No bank appears to have one yet, though banks' model risk managers have this responsibility.
Changing nature of daily work
Banks are starting to unleash generative AI on all employees. Last fall, JPMorgan Chase gave more than 200,000 of its employees access to LLM Suite, its internal platform of generative AI tools. It aims to give every employee their own AI assistant, tailored to their specific job responsibilities.
Citi recently released two internal tools across its workforce of 140,000: Citi Stylus, which can summarize and search reports, and Citi Assist, which searches internal bank policies and procedures.
"Giving everybody ChatGPT is kind of the analogy of Excel in 1980, saying, look, we know this thing's going to change us," Abbott said. "Let's give it to everybody and let them figure it out. It's the power of the crowd that's going to win this one, not the power of the Ph.D."
Generative AI is "automating most if not all routine tasks and helping virtually everyone in the organization work more productively, accurately and effectively," stated an Accenture trends report released Thursday.
"As most of the repetitive administrative functions are removed from people's roles, and as AI becomes their indispensable tool, so the nature of work in the average bank will change," the Accenture report said. "Bank employees will spend more time using their essentially human skills: judgement, creativity, empathy and relationship-building."
Banks continue to give AI copilots to employees to help them with specific kinds of work. For instance, Citi gave GitHub Copilot to 1,000 of its software developers last year, to help them generate new code based on existing code.
"We're starting to see the productivity gains at scale in terms of what co-pilot technologies can bring," said Arvind Purushotham, head of Citi Ventures. "We're also looking at how we can apply generative AI to the compliance space and reg tech space and all other aspects of working in a financial services institution. If you think about AI and gen AI specifically, I think there's extremely broad applicability for that technology within financial services. So we're very deliberately thinking about what that intersection looks like and how to do it in a way that is safe, sound, and compliant."
Research that Evident released on Thursday found that the 50 banks in the Evident AI Index launched 138 AI use cases in the past year, half involving generative AI. About 86% of the gen AI use cases were for employees only; the other 14% were client-facing tools like State Street's Alpha Platform.
Some use cases involve customer engagement, for instance, chatbots that answer questions. Others focus on knowledge access, for instance, summarizing documents for financial analysts.
In product development, a lot of work goes into testing, Abbott notes. Many banks have a shortage of product engineers, but a huge backlog of work, he said.
"My hypothesis is that AI will take on a lot of the mundane work, allowing banks to get to the backlog of things they still want to do," Abbott said. "The nature of work that we perform changes dramatically."
In banks' call centers, generative AI is making customer service representatives more efficient, not only by finding answers to questions but also by automatically generating summaries of calls. These automated summaries save about 15% to 20% of reps' time, freeing them to spend more time talking to high-value customers, Abbott said.
"Clearly, you could take out jobs in the call center, but banks don't want to," he said. "They want to use that time instead to figure out how to cross sell and upsell."
In commercial lending, generative AI can gather needed documents, extract the right information and put it in the required format. This could free a commercial loan officer up to look at 20 loans rather than 10, Abbott said.
"Everyone sees the knee-jerk reaction of, it's going to eliminate jobs, but the real value is in creating the capacity to generate more revenue," Abbott said. "The best banks in the world are looking at saying, how do I use that time to take in more loans, evaluate more opportunities."
