Author: Raabiya Singh
Scienaptic’s presence at CUNA GAC 2022 was hard to miss. The leading AI based credit underwriting platform provider organized an Industry Trends Panel session including Jennifer O Callaghan (SVP, Numerica Credit Union), Pankaj Kulshreshtha (CEO, Scienaptic), Soloman Semere (Senior Director, Lexis Nexis), Floyd Rummell III (CEO, Northern Hills Federal Credit Union) and Richard Waddle (CMO, CFO, CLO at GESA Credit Union). In case you missed the live session, here are the highlights and key takeaways from the panel discussion. We've summarized the panelists' answers to the four big questions we had on the potential of AI for driving financial wellbeing and enhancing member experience for credit unions and included some of the most memorable quotes.
What challenges are credit unions facing in competition with big banks? What features and products are being offered by large banks that credit unions can add to their product mix?
Evident challenges credit unions have experienced in competition with large banks is the branding of credit unions and banks through the years. Big banks have bigger budgets for advertising that credit unions don’t. It is a misconception that credit unions lack technology as they have every single technology that big banks do.
Despite the disparity in budgets panelists agreed that many big banks did not live up to consumer expectations. Today credit unions using AI powered loan decisions are taking pride in offering right product/service at the right number at the right time. Credit unions also focus on financial education such as Home Buying 101, understanding a credit score, tools to save for a down payment so that members are in a better position to make a good choice.
Credit unions also have an advantage in the data space. Many larger institutions don’t have a long-term relationship or history with the customer. But historically members have an affinity to their credit unions and trust them with their information. Today even younger consumers are looking at banking in a credible, trustworthy way.
“For us, AI has brought speed and convenience of the transaction. Our staff can now use their time and expertise in financial education and finding out if a member wants to own a home, start a college fund for their family or start a retirement account.” - Floyd Rummell III (CEO, Northern Hills Federal Credit Union)
“AI provides many opportunities to look at different ways of how we can help our members. I mean, ultimately, that’s what we are all supposed to do, right? That’s the beauty of being a credit union, looking for ways to serve our members better and deepen those relationships. It gives us more flexibility and more time to spend with our members and allows us to look at what they might need outside of what they’re asking.” - Jennifer O Callaghan (SVP, Numerica Credit Union)
Credit unions today have access to more data about their members through power of local intimate relationships. More data combined with AI can help them engage with them more proactively. What AI programs is your credit union is running?
Panelists agreed that AI based lending decisions was the core program their credit unions were focusing on. The other pressing area where panelists wanted to use AI was across the customer life cycle for pre-qualifications, pre-approved offers, ad-buying, predictive analytics, and digital marketing. Credit Unions want to personalize service at every touchpoint and want the ability to make prompt, instantaneous decisions for their members.
Adding to the first point on using AI for lending, Richard Waddle, CMO, CLO of Gesa Credit Union mentioned that one of the biggest opportunities for their credit union is that they excel in credit underwriting with AI. Credit score is not a good indicator of creditworthiness for marginalized and underserved communities. So, taking in data from alternate data providers like Lexis Nexis and other information on how members manage their account goes beyond just a few indicators of credit. One may have fifteen indicators up to a thousand indicators on which credit decisions are based, no human underwriter can go through all those and decide, but AI can. AI has the power to make instant decisions and help credit unions better serve their underserved communities.
Jennifer O Callaghan, SVP of Numerica Credit union shared that what is exciting about using AI for underwriting loans is the idea of access. There is significant need of credit for people who are unbanked, under-banked or considered un- creditworthy by big banks. This is a huge opportunity for credit unions. Credit unions are willing to be more creative, more thoughtful about the person instead of just the credit score. And the AI for loan underwriting gives them more flexibility with more data points to fill those credit gaps to provide those people a chance which can be life-changing!
As a society, we need more credit because disruptions are happening more regularly. But statistics suggest a different picture. 80% of people in the US have never been delinquent, but only about 40 to 50% of people get credit, which means that there is a massive opportunity to lend. What are the big challenges you see in deploying data and AI initiatives in the organizations? And how are you planning to overcome them?
The biggest challenge the panelists faced in deploying AI initiatives was getting staff buy-in and having them understand the process and objectives of using AI. AI is not to eliminate a single employee’s position, but it is a tool to enhance their ability to serve the members. Elaborating on that point panelists agreed that a human underwriter’s experience is valuable and should be used in a targeted and focused manner rather than using them to process applications in large volumes.
Another pressing issue discussed was setting up a use case for AI because there are so many possibilities. However, credit unions can dip their toe in the water and prioritize, test, and refine the model one step at a time. It is essential that the organization has a process to continue refining so that they have more confidence in the model as they expand to different portfolios or use-cases.
“At first, it was a little overwhelming, but we’ve got a really good process, especially working with the Scienaptic team. We change a couple of rules, test them, verify them, wait a couple of months, make some more tweaks, and just go and take it one step at a time because there are so many possibilities like you mentioned. This process makes sure we have confidence in the model and then we can move forward to continue to expand the credit that we’re able to underwrite.” - Richard Waddle (CMO, CFO, CLO at GESA Credit Union)
“AI and machine learning have the reputation of being a black box which is something that sophisticated banks have difficulty wrapping their head around. Companies like Scienaptic understand and navigate those waters very well. You can see the results and see all the benefits.” - Richard Waddle (CMO, CFO, CLO at GESA Credit Union)
It seems like AI and automation are table stakes for credit unions now. So what is the next phase of AI and what are the next steps?
Similar to earlier conversations, panelists circled back to discussing that giving loans to A+ and A members is easy. But credit unions need to make faster, better decisions and disburse more loans to more members. These are low-yielding loans and in today’s competitive environment, credit unions need to make loans to more challenged members.
The next phase could also include bringing in more factors into a credit decision along with testing and refining models so credit unions can help members with a lower credit score. Offering personalized financial services to members is something credit unions can always work on.
In contrast, members who may have a little bit less of a relationship or a little less trust, we can protect ourselves if there’s a fraud score. Instead of setting one limit across the membership, we can personalize the app based on our relationship with the member. And I think that’s something we’re working on right now and something that’ll prove to be an important part of AI coming forward.
AI should be about decision support rather than the decision-making itself. There’s a little bit of fear that we get out of what AI will do to us, humans. The reality is the combination works well. For example, the best chess algorithms found that putting a chess player with mediocre AI is better than strong AI alone. We should aspire to create a world where there is strong decision support and human beings learn to work with this multi-dimensional, fast-changing information because we, as humans, can’t process the information quickly enough. However, despite all proof of AI doing wonders in lending decisions, banks have not put it into production. The good news here is that the credit unions are significantly ahead at adopting very quickly and AI is helping them become supreme leaders of the lending industry.
Success Story “One day my direct manager came to me and said, this is a loan that the system (Scienaptic) said approve but review. So, we reviewed it, and she said I don’t know what took that long. We looked at the loan, and it was a son who had a very challenged credit, but the father was going to co-sign, they had a down payment and the loan value on the used vehicle was actually pretty good. As credit union leaders we need to make sure that the staff knows that these are loans that we need to do all day long. And in the past, without AI, that loan would never have even got looked at. But since it said approved but reviewed, it got looked at, and we could grant that loan and get this young man a vehicle. I’m sure that in the long run, you all know once you help a member out, especially ones that are credit challenged, they come back forever. You’ll have them forever, and I think we will have a credit member with this young man forever. So that’s one of my stories that I wanted to get to everybody, one of our early success stories we had, and I think it’s one of the many that we’re going to have.” - Floyd Rummell III (CEO, Northern Hills Federal Credit Union)
*** END ***
For more details about the platform visit: www.scienaptic.ai
Comments