Transcript:
Penny Crosman (00:03):
Welcome to the American Banker Podcast. I am Penny Crosman. In some organizations that deploy AI, there are holdouts: People who don’t trust AI models and who would rather do their own work using their own brains. But banks, credit unions, and others are under pressure to use AI to become more efficient and to prove to their boards, customers and investors that they’re using modern technology. One chief information officer navigating this dilemma is Raymond George at Clearview Federal Credit Union in Moon Township, Pennsylvania. And he’s with us today to share some of his experiences and thoughts. Welcome, Raymond.
Raymond George (00:38):
Thanks for having me.
Penny Crosman (00:39):
Thanks for coming. Can you tell us a little bit about your credit union? What is its field of membership? How many members do you have? What kinds of neighborhoods are you serving?
Raymond George (00:50):
Sure. We are located in Pittsburgh, Pennsylvania. We service the 10 counties around Pittsburgh. We have about 140,000 members. We are a community charter. We were formerly US Airways Federal Credit Union, but in 2004, they converted to a community charter. And so we’ve been community since that point, and here we are.
Penny Crosman (01:15):
Do you have a lot of pilots and flight attendants and such as members?
Raymond George (01:19):
Yeah, that’s our legacy membership. So they’re all around the world, but the Southwestern PA membership is obviously where we grow a new membership from. The legacy airline can still have family members and such join, so we have served both sides there.
Penny Crosman (01:40):
So I kind of kicked this off talking about the pressure that a lot of companies are under to do more with less using AI. Do you feel that pressure and do you feel pressure to keep up with what the big banks like JPMorganChase and such are doing?
Raymond George (01:55):
Well, yes. Doing more with less isn’t new. I mean, that’s something that’s always people have been striving for a long time, but the advent of AI, it’s a little bit of a game changer and large banks are using that to gain efficiencies all across, whether it’s financial services or all industries really, are trying to do the same thing, leverage AI for efficiencies because it does lend itself to be very good in that space as far as personal productivity, as well as corporate functions to be more efficient across the board.
Penny Crosman (02:37):
So as we’re taping this, Jack Dorsey at Block has just announced he’s laying off 4,000 people from his company due to gains in AI-related productivity. In a shareholder letter, he wrote, “Intelligence tools have changed what it means to build and run a company. We’re already seeing it internally. A significantly smaller team using the tools we’re building could do more and do it better. And intelligence tool capabilities are compounding faster every week. I don’t think we’re early to this realization. I think most companies are late. Within the next year, I believe the majority of companies will reach the same conclusion and make similar structural changes. I’d rather get there honestly and on our own terms than be forced into it reactively.” What did you think about this? And do you agree that it’s kind of the inevitable future for most companies to have to drastically lay off large numbers of people due to the efficiency gains of AI?
Raymond George (03:38):
Well, that was certainly a shocking number in almost half of the organization is being laid off. AI is powerful. Is it that powerful? That’s a good question. But from a realistic standpoint, will companies do that? Yeah, there are efficiencies to be gained in AI that could lead to job cuts and streamlining of products and services and how they’re delivered. But I can’t comment on their business model and how they can make such a drastic move that quickly. It’s not something we’re planning here at ClearView, obviously, but to the point of, will A, I impact most jobs and make things more efficient and can you do more with less going forward? Yes, we’re in a growth period here at Clearview. And so our goal is to obviously grow our credit union, our membership, and our offerings to our membership without necessarily growing with a number of people to keep up with that demand and those services.
(04:39):
So is there efficiency? Yes. Laying off half your organization, you’d have to look at that business model and figure out if that’s right for you or not.
Penny Crosman (04:47):
A fair answer. So given that we are seeing these kinds of layoffs, and to be fair, some of these kinds of layoffs have been announced, and then we’ve seen a big rehiring at companies like Klarna who had to go back and hire a bunch of new people. How do you get people motivated to want to deploy and use AI knowing that there are a lot of people losing their jobs because of the technology?
Raymond George (05:18):
Well, we’ve been trying to communicate the realization that AI is going to change or impact most everybody’s jobs. And what we’ve been trying to do at Clearview is get people trained on the use of AI, how it’s to be used so we can start that changing of processes and job functions moving forward. Again, depending on what your business model is and what you’re doing, our intent is to grow without necessarily growing same amount of staff that we have today. We want to double our membership, but we don’t want to double our employee group. We want to be efficient. But what we said, if AI may not take your job at somebody who knows AI will, and that’s a very important statement and we keep trying to hammer home. If you like, “Well, Clearview’s going to make me use AI, I’m leaving.” Well, where are you going to go that’s not using AI?
(06:12):
There’s not a whole lot of places you’re going to go that aren’t using AI. So it’s really just that transformation of the job that this technology is bringing.
Penny Crosman (06:22):
Yeah. And it seems like you take a pretty strong stance. I saw that you recently said, “If I get you an AI license and you’re not saving the amount of time that the license costs, then you’re not using it right.” When AI is not useful or produces errors, is it always the user’s fault?
Raymond George (06:38):
Well, one of the other things we tell people as a human, you got to bring something to the table. So part of using AI is also checking AI and making sure it’s not giving you inaccurate information because it can. It’s part of the tool and those tools will get better in each revision and gets better and they’re training models on specific tasks and it makes them better. But at the end of the day, we’re not turning over people’s jobs completely to AI and removing the human judgment and ability for somebody to make sure that that information’s correct and used the right way. AI is a great tool, but it’s a partnership with the human using the tool that makes it right. We have done training on doing prompting and things like that so we can make sure you aren’t getting the best information out of the tools possible.
(07:32):
But yeah, can it give you some bad things? Sure, but hopefully our employees are sharp enough to pick up on that and make sure that that’s not used.
Penny Crosman (07:42):
I’ve heard recently of companies monitoring employees’ use of AI and at least thinking about making a part of the job requirement and part of layoffs of people who are holdouts and don’t want to use it. Would you go as far as that?
Raymond George (08:00):
Well, here’s how this is going to play out, in my opinion. As AI becomes more used in the work environment, AI and the uses of AI are going to start to get ingrained into the actual job function of whatever that job is. And that’ll just be the way you do your job. It’s not going to be, well, I don’t want to use AI. It’s going to be, this is the way you need to do your job. Just like anything else, any upgrade. If you were a biplane pilot and you got a new jet, you had to learn how to use a new jet and yet I go, I don’t want to … Well, that’s just part of the job now. There’s other things that need to happen. So while AI is this quote, unquote, semi-new thing that’s out there and I’m going to use AI for my job, AI is just going to be part of the job function.
(08:47):
This is going to be part of the fabric of how people do things.
Penny Crosman (08:52):
That makes sense. Well, there’s also been a lot of talk lately about the SaaS apocalypse that traditional software or software as a service is going to be replaced by AI agents that can be assembled or put together to do what software used to do. Do you see that really happening? Do you see that replacement taking place?
Raymond George (09:20):
That’s maybe not in those terms. Is it going to have an impact on it? Oh, absolutely. It’s going to have an impact on many things across the board. Will it replace it or will it mutate it or just become part of how it’s delivered? Probably. I don’t know if it’s the end of software as a service, but I think it’s going to have a big impact to it. It’s going to be the delivery channel. There’s been delivery channels changing all the time. You went from Netflix renting videos to streaming. People still watch videos. The delivery system just changed.
Penny Crosman (09:57):
That’s true. And what are some of the ways that your credit union uses AI today?
Raymond George (10:03):
Well, we did some training on prompting, and so we have AI champions across the organization. So it’s used from basic HR things, job functions, job descriptions to our BI team uses it to analyze files and looks for trends, generate formulas for Excel or Power BI. Our coding team uses it to generate some code. Our call center manager can look at chat logs and look for themes that might be happening within the user base that they can get ahead of and train their other people on. It’s just, we use it for scheduling. Our branch personnel, there’s a number of things that across the board. We also have a lending piece that they analyze portfolio, analyze delinquency, analyze fraud. There’s a number of things that we use it for.
Penny Crosman (11:07):
Can you share some of the specific models that
Raymond George (11:10):
You use? Sure. We have a corporate license for ChatGPT. We also have Copilot, and there’s another product called Lulu, which is from Zest, which is geared towards lending and Credit Union 5300 reports, which if you ever had to do any research on your peer group or jive through 5,300 reports, this thing is a massive, and I mean massive time saver, to be able to get through that, so just to do analysis on how you’re faring, how your portfolio is faring and things like that. So yeah, those are the three main models we have, but we’re obviously looking at all of them on what might be good for what function.
Penny Crosman (11:47):
Do you also look, and this might be kind of a tangent, but there’s been a bit of a storm brewing with AI companies and the Pentagon, and some are willing to work and do exactly what the Department of War wants, and some are not. Do you care about things like that, those sort of macro issues that the major AI companies are dealing with?
Raymond George (12:13):
Well, sure. I mean, they’re going to filter down into the Pentagon wants to use it for certain functions. You’re going to have security companies want to use it for certain functions. You’re going to have financial institutions use it. And where do you draw the line on what’s ethical use of the technology and what’s best for things? You could even get into predictive analytics. And if I analyze all your transactions and I think that maybe I see you shopping for baby clothes, should I start sending you ads and promotions for those types of products? And other people might see that in your house that you may not necessarily want them to see that in your house. So where these AI companies draw the line on FA, I’m watching it because it’s going to filter down to all uses of AI really.
Penny Crosman (13:16):
That makes sense. And how do you try to measure the outcomes? Are you looking for specific measures of efficiency? Are you looking for a certain amount of return? Are you looking for revenue in some cases? How do you look at that?
Raymond George (13:33):
Well, a lot of it has … We also have a big NRPA department that is getting into automations using Agentic AI, and that one’s an easy group because we measure the number of hours saved in a year, but we’re also looking at whenever we get into looking at models for maybe credit scoring and things like that, we’ll put parameters around what we’re looking for to make sure it’s useful and we’re getting our money back or seeing a return on that investment. But those are things that we’re starting to get into now other than the saved hours. We’ve been going after that one now for a couple years. That’s pretty successful. But as those processes start to converge and you get agentic AI agents out there, that’s going to change that game a little bit too.
Penny Crosman (14:32):
It’s interesting because I remember covering RPA about 10 years ago, there was this big kind of boom in RPA. And I remember seeing this analyst saying that all cubicle jobs or most cubicle jobs were going to go away because RPA was just going to automate everything. And then it didn’t turn out exactly as predicted. In fact, a lot of these projects kind of died out, I believe partly because these bots were vulnerable to UI changes, application updates, changes in process, et cetera, that then made them useless basically. Is it different this time? And do you think that Agentic AI gives RPA the juice it needs to really work?
Raymond George (15:27):
Well, anytime when RPA came out, you’re right, all the cubicle jobs are going to disappear, and they did. AI came out. All these, whether it’s coders or all these jobs are going to disappear, they’re not. So once you get past the crazy claims of any type of new technology, everybody, when something comes out, it’s always the extremes. That’s what’s in the newspapers, that’s what gets people to read things. But when you come back to what is it useful for, you’re right, screen scraping technology, somebody changes the website, then it blows up the RPA. Some of the tools are far better than that, what they used to be now. The Agentic is also giving them a little bit of an influx of energy because now it can do more than it could because it can adjust to the inputs that it’s getting or the items that it’s finding.
(16:28):
But at the end of the day, if it were that easy, nobody would have any jobs in the tech field by now because AI and RPA was going to replace everything.
Penny Crosman (16:39):
Right. So it’d be interesting to see if there is a similar hiccup in this wave that we’re seeing now. I
Raymond George (16:47):
Was going to say, you see this all the time, and that’s just part of the … There’s this hype that hits that everything’s going to change. And then when it settles into reality, I think you even see that with some of the tech companies, all the coders and all these things are going to be gone. And then what was it? Last year, they said 90% of AI initiatives fail. It’s because maybe you were trying to do something that wasn’t ready to do yet, but at the end of the day, you don’t throw the incredible gains you can get from AI out because of 90% of ginormous AI projects fail.
Penny Crosman (17:30):
All right, that’s well put. Well, Raymond George, thanks so much for joining us today. And to all of you, thank you for listening to the American Banker Podcast. I produced this episode with audio production by Adnan Khan, WenWyst Jeanmary and Anna Mints. Special thanks this week to Raymond George at Clearview Federal Credit Union. Rate us, review us and subscribe to our content at www.americanbanker.com/subscribe. For American Banker, I’m Penny Crosman, and thanks for listening.