Did you blink?
A year ago, the banking industry was rife with skepticism about the promises made by backers of GenAI.
“People were trying to figure out, is this hype? Is it real? What’s there?,” says Keri Smith, global banking data and AI lead at Accenture. “Fast forward to today — there’s been a lot of proven experimentation as well as new processes in production and many public examples to learn from. Most banks have come to understand that the technology is real.”
Does that mean that bankers who did not immediately jump aboard the GenAI train have fallen permanently behind, having not taken the tech leap of faith? No, says Smith — not yet.
“We strongly encourage our clients and our partners to put a foot in the water, get started. You can’t sit on the sidelines,” says Smith, in an interview with The Financial Brand. “There is ‘first mover’ advantage here, but it’s not too late.” A major challenge, she adds, is not expecting perfection on Day 1 from GenAI, but instead anticipating gradual improvements in whatever the bank applies it to.
One of the major shifts over the last year is how and where GenAI can be applied. Not so long ago, many who were seriously looking at GenAI figured, “large language model, eh?” and eyed functions like bank marketing as a good fit.
Now many bankers believe that GenAI will have impacts throughout banking organizations, according to Smith.
In Accenture’s report, “The Age of AI: Banking’s New Reality,” 32 banking functions are identified as having potentially high impact from generative artificial intelligence, and aspects of marketing account for only three — lead origination, new offer management and digital content creation and management. Also on the 32-item list: business banking relationship management, wealth management advisory services, branch advice, client onboarding, document and knowledge management, anti-fraud management, legal services and risk management.
“I’ve been in this space for over 20 years, but this is the first emerging technology I’ve seen that has had such a rapid pace of innovation and adoption,” says Smith. Bankers, even in the C suite, need to constantly renew their AI literacy, such is the evolutionary momentum.
Even more important, in Smith’s view, will be how banks think strategically about how they operate going forward. “GenAI is a chance to reimagine the way of working that we want,” says Smith. “You don’t want to copy and paste broken processes and then just add technology on top of that.”
Smith discussed how institutions should approach GenAI adoption, how newcomers can find a way to begin, how to apply risk management to GenAI, and how to keep GenAI customer friendly when what many hear about is mostly GenAI “hallucinations.”
This Credit Union Staffed Nine Branches With Just Three Employees.
Needing to improve staff efficiency, Great River deployed new technology to centralize staff. The results? An 80% decrease in lobby wait times and 4-to-1 FTE.
Read More about This Credit Union Staffed Nine Branches With Just Three Employees.
Transform Your Credit Union’s Indirect Lending Business with DPA
Discover how document processing automation (DPA) uses AI to streamline lending processes, enhancing efficiency and accuracy, while ensuring compliance and protecting sensitive information.
Read More about Transform Your Credit Union’s Indirect Lending Business with DPA
How to Move Down the GenAI Path in Your Bank
Smith works chiefly with large banking institutions, both domestically and internationally, as well as regional players. Some remain in the experimentation stage, often trying out ideas in back-office functions before they tinker with processes the public sees. Others have started to put some GenAI into actual production.
A key lesson learned is that a proof of concept can’t be intelligently scaled unless the bank has thought past the experimental stage. “It’s important that bankers understand what the overall vision is for GenAI in their institution, versus just trying out piecemeal experiments and expecting to scale them individually,” says Smith.
Smith explains that this means more than drawing up a map or a to do-list. A key issue: Understanding what improvement to a existing process will actually look like. Metrics must be set for three months out, six months out, a year out.
“I’m not going to get 100% improvement immediately,” says Smith, “but if I’m able to get 10% better, 20% better, that’s good.”
Another key decision is what the bank hopes to improve with GenAI — is it going for cost control, for instance? Smith says most institutions have put a priority on cost savings over other goals.
Institutions that have started experimenting with GenAI should maintain multiple projects simultaneously, Smith suggests. “You’re not going to be able to tackle everything at once, but you’re able to do things in parallel.”
Read more: Why it’s Time for Banks to Hire a Chief AI Officer — and What That Looks Like
What to Do If You Aren’t Even on the GenAI On Ramp
For banks that have done little or nothing with GenAI, knowing how and where to start may seem overwhelming.
Organizationally, a priority is setting up a generative artificial intelligence governing group, a GenAI “SWAT team,” in Smith’s words. A key member of that group will be someone from the risk management team who understands the bank’s model risk management policies and practices. Remember, at bottom GenAI is a computer model, and adoption of such tools is already subject to regulation.
In fact, Smith sees this as an advantage for the industry, in that careful analysis of outside models should be baked into bank decision-making already. “GenAI is building on the back of muscles that the banks have already developed.”
Merely studying GenAI isn’t the group’s role. Smith says a key early task is setting secure guardrails for proofs of concept, as are “secure sandboxes,” protected virtual areas where approved experiments can be conducted. Safeguarding bank and customer data depends on such approaches.
Smith says bank staff must have the opportunity to start working directly with GenAI in order to assess its possibilities and to see how well it works with the bank’s existing usage of artificial intelligence.
“There’s formal training that people can take, but there’s nothing like hands-on, safe experimentation,” says Smith.
Read more:
- The Speed of the GenAI Revolution Is Creating a Regulatory Vacuum. What’s the Risk for Banks?
- In The Wake of Google Gemini’s Chatbot Debacle, An Object Lesson for Banks
- Your Marketing Team Needs New Skills to Succeed Today. Here’s How to Implement Them
When GenAI Crosses the Threshold of Your Front Door
At some point institutions will bite the bullet and let GenAI interact with customers or with their live data. Smith believes it will be important to alert customers when they are offered a product or process in which generative artificial intelligence plays a role. Both staff and the public must be clear on this and have confidence in what the bank discloses.
Some consumer paranoia exists about GenAI, but Smith says Accenture research indicates that many customers will accept its use with their data — if they feel that they are receiving some benefit in return.
Asked for an example, Smith points to a frequent beef about bank customer service — having to explain a situation all over again when trying to work out a problem or address a need and being transferred from one staffer to another.
“I shouldn’t have to educate you on what happened two days ago,” says Smith. “I want to walk in and find that you already know.” GenAI can help with this type of situation.
One last pointer from Smith concerns the customer experience of using processes controlled by GenAI. She says that some customer-facing GenAI provides inferior look and feel, which can become a friction point.
Smith says this is common enough that Accenture has brought in user interface and user experience experts on some assignments.