Ask a credit union c-suiter if their institution is implementing artificial intelligence and the fast answer is, you bet.
But quietly ask, how is it going? And probably you will hear, there are no deliverables worth mentioning.
Call that expensive tinkering but it is not unique to credit unions. MIT recently analyzed AI deployments at some 300 publicly held companies and concluded that 95% produced nothing that impacted the bottomline.
But then there is what credit union consultant Kirk Drake is experiencing in his work with institutions and, says the CEO of CU 2.0, many credit unions already are showing tangible, positive results with bottomline impacts.
Drake, understand, is an AI veteran. He’s the author of FINANCIAL, a 2020 book that explored the early days of AI in credit unions. And he’s definitely kept up with the field’s rapid evolution as large language models birthed new, powerful AI tools such as ChatGPT, Google’s Gemini, and Anthropic’s Claude.
In an hour long conversation, Drake vividly described where AI is in today’s credit unions and, importantly, what smart credit unions are doing to position themselves to be AI winners.
One reality: just as the advent of the personal computer in the 1980s dramatically reshaped work flows in all companies, credit unions included, so AI will reshape the workplace but it will happen at a much faster rate because this is change greased by the flow of many billions of dollars into the development of AI tools.
Drake talks about various phases of AI in credit unions. A first phase “is the one that got everybody all hyped up, which is, I can use it to do my email writing. People got pretty excited about that and have seen decent gains,” he said. Go ahead: ask ChatGPT to read and summarize your email and suggest responses. You may find that the hour or two you now devote to email can be cut to a half hour or less per day.
A second phase, said Drake, is where AI is asked to clean up a credit union’s policies and procedures—for instance, the documents that guide call center employees as they respond to member questions. Drake related that exactly this project was implemented at One Nevada Credit Union where AI tools developed by Senso have enabled dramatic improvements in how fast employees can accurately answer member questions, mainly because Senso AI has sorted through the many, many pages of instructions, distilled them down to their essence and made the result readily searchable. In some cases, incidentally, AI found documents that offered contradictory advice to members and of course that was cleaned up. The result has been happier employees, happier members, and this document clean up is a project most credit unions would benefit from, said Drake.
In a similar vein, Drake pointed to fintech Casap, whose AI tools automate and simplify the dispute/chargeback process in credit unions. Chargebacks total more than $60 billion annually and yet a lot of disputed charges are believed to be fraudulent. “Casap has built a completely autonomous dispute resolution process,” said Drake—and that means many credit union employees are freed from work many found frustrating because AI tools are doing all the work.
Eltropy, meantime, has released its AI powered Collections 2.0 which seeks to automate the collections process and credit unions that have implemented the tool say it pays for itself in increased collections.
The takeaway: narrowly focused AI, says Drake, is often producing the best and quickest results in credit unions.
But ideas can get bigger. At CU 2.0 Drake says “we’re working with some credit unions on building a series of ChatGPTs that optimize core deposit generation through marketing tactics, data analytics, and building automation into the process around consumer engagement. Better operating efficiencies increasingly are critical to credit unions as they compete with big banks and fintechs and AI is showing the way," said Drake. That is as big as credit union ideas get and AI is centerstage in driving results.
Where do you want to dig in with AI in your credit union—something small, something big?
Drake acknowledges most credit union execs have a couple dozen ideas about where to put AI to use but, honestly, that probably will produce little value because efforts will be too diffuse. Narrow your focus to get better results. “If you tackle two or three ideas per quarter,” says Drake, “you are going to land on some that become a permanent part of your institution.”
One big question that has held back many credit unions from making aggressive use of AI tools is fear that the data they input into an AI tool suddenly is in the wild and out of the institution’s control. Drake says there are simple fixes. For instance: in ChatGPT go to settings and tell ChatGPT not to train on your input. Then take the next step of never using real names or PII in AI prompts. By all means, ask ChatGPT if this is a loan app that should be rejected but do not use the member’s name or any other PII. “Problem solved,” says Drake.
How can you know if any of your AI efforts are worth the bother? As you implement AI trials, says Drake, keep scorecards and be ready to move off projects that aren’t generating tangible results. There will be failures inside credit unions as they move aggressively to adopt AI tools. Expect that. But move off failures fast, says Drake.
There will be successes, too, lots of them, and the way to grab your share is to get started on AI trials now. Don’t delay. Early adopters are already reaping benefits. Get your share by taking the AI plunge now.
