AI Questions Answered for Credit Unions

Credit union ai questions and answers from cu 2.0

This piece is based off of a roundtable discussion at the CU 2.0 Brainstorm Event in July 2021. It is not intended to be comprehensive—rather, it will provide a cursory introduction to AI for credit unions.

Join our next Brainstorm Event here!

More and more credit unions are using artificial intelligence (AI) to power their technology. Whether your credit union is already on board, biding its time, or skeptical, know one thing:

AI will continue to change the way financial services are rendered for the foreseeable future.

Financial institutions that don’t use AI today will use it tomorrow. Or they won’t be around anymore. The same way that digital and cloud have taken over, so will AI. And it will happen faster.

Here are a few of the major topics, questions, and answers about AI for credit unions today:

Want to know more about credit union technologies and solutions? Join our free Fintech Call Program!


1.  What Is AI/ML?

We’ve all heard the terms, but what do AI and ML (machine learning) mean?

If you want a long, informative, and entertaining answer, buy this book.

If you want the short and dirty definition, it’s this:

AI and ML describe computer algorithms that solve problems… and teach themselves to solve those problems even better with more time and data.


2.  Which Problems Are Best Solved by AI?

This is a difficult question to answer, because the question itself is flawed. A better question might be:

“Which problems do credit unions need to solve, and can AI help with any of those solutions?”

The fact is that AI itself isn’t going to solve problems. Rather, AI will optimize solutions to those problems, increasing the efficiency of older, less sophisticated methods.

However, anything affected by an abundance of data may be an especially good target for an AI solution.


3.  What Are Some Common AI Use Cases?

Again, it’s worth noting that you should read the book for a better view of AI use cases in finance.

There are many common use cases for AI in finance. However, there are a handful that see more use than others at this early stage:

  • Chatbots: Natural Language Programming (NLP) is a way for machines to speak with humans conversationally… like humans. Chatbots can answer queries, provide direction, and even perform menial tasks, all without engaging call centers or service reps.
  • Fraud prevention: AI is incredibly adept at pattern recognition. So, when something looks out of the ordinary, AI will catch it before it becomes a problem. Similarly, AI is smart enough to know when some abnormal transactions are perfectly legitimate.
  • Credit decisioning: AI can pull from multiple data sources to make quick decisions on loans. Thus, it provides much faster loans at scale, increasing profitability. It also improves the borrower experience. Finally, AI is generally more accurate than traditional methods!
  • Document management: AI can scan documents more quickly and accurately than humans now. This provides incredible efficiency for onboarding, document collection, legal work, and more.

There are many more uses for AI in credit unions, but these are a handful of the most popular so far. Ultimately, the question won’t be about what is most common—rather, it will be about which uses cases make the most sense for you… and which are available!


How Can Credit Unions Assess AI Vendors?

This is a topic for an entire blog. However, the short version is this:

Ask yourself the following questions:

  • Does the AI solution solve a problem we/our members have? Don’t go into AI just to go into AI. Do it because it solves a specific, answerable need.
  • Do we have to train the algorithm ourselves, or will that be handled for us? If you don’t have data scientists on board, you’ll want the vendor to provide that service and customization for you.
  • Is your credit union ready for the AI solution? Unfortunately, integrations with existing tech stacks are often major blockers. Will you be able to integrate the solution without too much difficulty?

Additionally, beware any vendor that emphasizes its credentials more than its results. Some fintechs are ready to rest on their graduate school laurels… others are ready to show you proof that their solution works before presenting a proposal.


What Are the Biggest Obstacles to AI?

Fortunately, we’re getting past the point of bias being a key issue in AI. Data scientists know to look out for and exclude data or results that might not play fair.

However, there are still some obstacles. The biggest ones include:

  • Data: An abundance of good, structured data is key to AI success. Credit unions have a lot, but it’s still often insufficient to power accurate AI models.
  • Explainability: It’s not always good enough to have the right answer… sometimes you have to know how you got that right answer. Especially when it comes to regulations, explainability is a big deal. And it’s one that many AI vendors are still grappling with.
  • Trust and integration: Any AI solution has to fit with both the credit union’s tech stack and the people who maintain it. Both the culture and the existing tech has to be compatible with the AI solution if it’s going to work.

Of course, there are other obstacles as well. But generally, those obstacles will exist with any vendor, AI or not.


Next Steps for Credit Unions and AI

CU 2.0 works closely with countless AI vendors and visionaries. We also consult with credit unions and regulatory bodies to guide the industry through these unknown waters.

Join our Fintech Call Program to keep abreast of the latest in AI for credit unions.

And subscribe to our blog—we have a lot more AI content planned!


Recent Posts