By Ray Birch
MADISON, Wis.—Should credit unions be using AI to conduct focus group research?
One expert says that’s a good use for the tool, saying there are insights to be gained from the tool as a credit union asks questions of the technology just as a CU would a group of consumers or members.
“I see a number of very exciting opportunities. I'm sure all credit unions are thinking about the possibility of automating processes,” said Ignacio Luri, assistant professor of marketing at DePaul University. “I know many credit unions are implementing chat bots and automating customer support.”
Luri partnered with Noël Manushi, finance and marketing student at DePaul, on the Filene study titled Text-Generative AI and the Future for Credit Unions.
He pointed to an opportunity the believes may be underused by CUs: “Integrating AI with current analytics systems to extract insights. For example, member reviews, social media conversations and all sorts of data,” he said.
An Interesting Use-Case
What may be most interesting now with AI and credit unions, according to Luri, is the use of synthetic data.
Synthetic data is information that is artificially generated rather than produced by real-world events.
Luri said synthetic data “mimics” consumer behavior.
“What happens if you have AI create data and you use that as legitimate data to get insights?” he asked. “That raises a number of concerns, for good reasons—but maybe not so much right now as it will in a few years. Right now, there are a number of small companies and startups already offering services for synthetic data.”
‘Valuable’ in the Near Term
Luri said that in the near-term the use of synthetic data is valuable.
Ignacio Luri
“In the short and midterm there's going to be a really exciting opportunity to use this data for things like prototype brainstorming--maybe design of a website. Instead of, or in addition to, showing the site to customers, have AI respond to give tips and feedback,” he said.
In such a scenario, the credit unions would be posing questions to the AI--just as they would focus group participants—such as, “Do you like this credit card or that credit card? Which one do you prefer? And what do you care about when you're choosing a credit union?”
That scenario comes with a caveat, however, with Luri explaining there are potential trust issues.
“I see the use of synthetic data as a great opportunity to complement human insights,” he said. “I don't think we're at the point where we can rely just on synthetic data. A lot of this is experimental now, but I think it's exciting. I am using it as an academic and researcher. I'm starting to use it and I've been impressed by the extent to which it really does mimic human behavior.”
One ‘Concern’
Nevertheless, Luri also believes the use of synthetic data has a limited shelf life, explaioning that currently AI primarily learns from the data available on the Internet, which is largely created by humans. But as AI sees increasing usage, more information that’s created by AI will populate the web, and then it will become a matter of AI learning from AI.
All of that “concerns” him, Luri said.
“AI is very good today at mimicking human behavior because it is trained by humans. But what happens in three to five years when the Internet is contaminated by AI generated data? That could create all kinds of trouble. I think eventually we will need to reconsider our trust in AI as a complement to human data, but right now it does make sense to use it this way.”
Luri contended that if a credit union is not using AI as an analytics tool, it is late to the game. He added that AI can be a big help to credit unions as they seek to automate tasks that are time-intensive, such as digesting large amounts information and then providing summary data, such as tax or compliance tasks.
Keeping AI on a Leash
But, don’t give AI too much freedom, he cautioned.
“I recently saw a story about a major airline whose chat bot told a customer they could get a refund and then directed the person to the airline’s website,” Luri said. “But the actual website said they could not get the refund. So, the customer sued the airline.”
He explained the judge ruled for the consumer, since the chatbot was speaking on the airline’s behalf, despite the fact it developed a refund policy itself.
“So, AI is far from perfect,” Luri said.