The Promise and Peril on ChatGPT in Banking Part 2

Jim Marous (00:00):

Hello, and welcome to Banking Transformed, the top podcast in retail banking. I'm your host, Jim Marous, Founder and CEO of the Digital Banking Report and co-publisher of The Financial Brand.

Jim Marous (00:22):

With the rise of generative AI, ChatGPT and its increasing impact on various sectors, including banking, is crucial to examine how these advancements are transforming the landscape, both for the better and for the worse.

Jim Marous (00:36):

From exploring the impact of generative AI on customer experience and engagement to discussing the ethical considerations and regulatory challenges, it's more important than ever to understand the transformative power of AI in banking.

Jim Marous (00:51):

This is the second part of a podcast we've done with Brian Roemmele, President of Multiplex. Brian is equipping our listeners with the knowledge and understanding needed to embrace the promise of generative AI and ChatGPT while helping us to mitigate the risk.

Jim Marous (01:10):

If you have not listened to part one, be sure to go back and listen to part one while this is part two of the same podcast.

Jim Marous (01:16):

So, Brian, thanks for sticking around and doing part two of this very important podcast on generative AI. In part one, we discussed a lot around the experiences possible, and the opportunities that generative AI and ChatGPT provide as we interact with customers and build better experiences and better engagements.

Jim Marous (01:39):

So, Brian, in the first part of our interview, we talked a lot about the customer experience, how it can be enhanced. And it's interesting because when you look at the amount of data that can be collected and the dialogue that can take place, one thing that comes to mind is, is it possible that this dialogue, this identity, this place in the journey that is shared between ChatGPT, and the customer can also serve to help in the identity and customer authentication process?

Jim Marous (02:08):

We always talk about, how do we find a universal way of identifying and recognizing that this is really who we think it is? Can this actually be some of the input to determine the authentication process and identity of the consumer?

Brian Roemmele (02:25):

Jim, that is a brilliant question. Yes, and I would say that if we contain as many data points as possible on that client, we are going to get that much closer to 100% certainty that we would've had in physical banking, physical financial.

Brian Roemmele (02:45):

The most important thing to understand is as AI models know more about you, it gets extraordinarily difficult to fool those models into thinking you're somebody else. And that's vitally important because we haven't talked about the dark side, and dark side is real fakes.

Brian Roemmele (03:03):

And using AI to create a video of me, I could very well be AI. A year from now, it may be debatable whether we ever had this interview, it could have just all been made up by AI. And they got a likeness of me because I tend to use my same background and they threw it together, and there we go. The same is going to be true with everything, special official functions like finance.

Brian Roemmele (03:32):

I happen to believe that the multiple data point validation schemes are the best, and the more data points you have, the better. So, if you have a model already built with that particular person, asking them just a couple of questions and how they respond to them, can give you a really great insight on whether you're dealing with that person.

Brian Roemmele (03:56):

Now, I don't want to give it away, but I can ask somebody through AI, six questions, very unique; that unless somebody heard that person answer those questions, they'll usually be answered most definitely the same way, using generally the same vocabulary. Not the exact same words, but the vocabularies range.

Brian Roemmele (04:20):

Because we can tell word usage and vocabulary nomenclature that people are used to using, and it would be nearly impossible for somebody to determine. And I think that's going to be part of future validation. I'm hoping that to have a universal ID system for a lot of reasons, I think there's a significant downside to that.

Brian Roemmele (04:45):

I think this will actually do better than a universal ID system. Because if we rely upon a universal ID system that is absolutely fallible (because they all are), we're going to rely upon something that ultimately will cause more harm than good at some point in time.

Brian Roemmele (05:02):

We all know this; at some point, all encryption is going to be cracked. Now, that's a quantum computer scenario. And so, the overly reliance on some sort of encryption to me is a fool's mission. It's going to be cracked, there will be no secrets from that point backwards.

Brian Roemmele (05:18):

Now, that point forward, you would have quantum-resistant maybe technology, but everything from that point backwards is now everybody's information, it's cracked. So, that's final.

Jim Marous (05:30):

Well, it's interesting, in talking to David Birch, who you and I both are familiar with in the recent past on a podcast we had, he really has moved away from identity and moved much more towards authentication, which it sounds like just wordsmithing.

Jim Marous (05:46):

But the reality is what you just brought up, the ability to understand the context in which people communicate, the way in which people communicate cannot be duplicated. Certainly, can't be be duplicated.

Brian Roemmele (05:57):

David is brilliant, and I noticed listening because I listen to you all the time - that shift is remarkable because we all had to go through it. This is the beauty of technology, is it's humbling. It's like whatever we thought a few weeks ago, my gosh, I have to change my mind. And that's part of never really getting stuck.

Brian Roemmele (06:18):

My entire life mission is to ... I want to constantly change my mind. I want to constantly be wrong because the wrongs allow me to get to the rights.

Brian Roemmele (06:27):

And so, we need to move past the idea of one monolithic way, whether that not even be DNA. I mean, people think, "Oh, that's going to be the ultimate identification," it isn't. It isn't, there's ways to simulate DNA to a level that I don't really want to speak about publicly. So, don't count on DNA. Oh, well, fingerprints, we know that's wrong.

Brian Roemmele (06:49):

I mean, it's going to be a multiplex, a montage of things that we're going to use to more understand who we're dealing with. And the certainty levels will actually be higher and as close to physical representation of that person in your building as it ever has been since the electronic age has started.

Jim Marous (07:11):

So, it's interesting, in the first part of this interview in the previous podcast, we talked a little bit about the regulatory challenges and the governmental issues, and you mentioned in fact that the concern rate now is drawing too close of a parallel to saying, "I want no risk as opposed to modified risk and how you look at this."

Jim Marous (07:30):

But what regulatory challenges and guidelines currently exist for the integration of AI in the banking industries, particularly in respect to generative AI and ChatGPT?

Brian Roemmele (07:42):

Wow, so this is more of a perception than actual law. The perception is how much information are you allowed to give AI? Well, if it's a big monolithic AI and you're mixing everybody's information, you have personal information, we know already that that's a problem, can't do it.

Brian Roemmele (08:04):

However, if you segregate that AI and sanctify in its own container and you've made every attempt to make sure that that AI is not mixed in general - now, outputs for the AI can be used, that's prompting the AI and say, "What does this customer need?"

Brian Roemmele (08:24):

Like the company AI can have a simple prompt to say, "What does this customer need today?" And the customer is notified. Somebody has prompt your AI at the company, ask this question; maybe even give them the ability to vote it down. Say, "No, I don't want that question asked," fine, move on.

Brian Roemmele (08:42):

Giving that level of control is unusual for a major corporation. It's almost like, "Is this guy really the hippie in California is talking about because we don't work that way." You are going to have to work that way. You don't have a choice because the regulatory environment is going to force it to a level where you couldn't even do that.

Brian Roemmele (09:03):

So, if you do it on your own, if you say "Everything's transparent, we are the transparency company when it comes to your data, but we protect your data, and here's how it's stored. Your AI is only available to you and only available to the company. And every time it's ever accessed, you are notified and you get first right of refusal."

Brian Roemmele (09:23):

Now, first right of refusal may go too far - I don't think so. I think this makes a person feel like, "Wow, this company caress about me." Nobody's going to ask a question, I don't want it to be answered. Because it becomes a proxy for that individual. The more it knows about you, the more it becomes a proxy.

Brian Roemmele (09:39):

But you've granted the right for this company to hold in protection and value the proxy of you in their company and that you are in control of it. And so, if you do it that way ... I'm not a lawyer, I've worked with a lot of lawyers on this, and this is not new to me. So, I'm not on the fly making this up. If you are doing it to this level, you're not currently violating any banking rules or laws.

Jim Marous (10:08):

No, in fact, you're actually going more towards their customer control than we do now. Because we try to think on behalf of the customer now. What you're actually saying is we're going to be completely transparent.

Jim Marous (10:20):

We'd like to use a tool to help understand you better, it may ask you questions. Oh, by the way, you can turn it on or off at your own will. And that ability, that control issue, is really the ultimate holy grail of what the government usually wants it to be.

Jim Marous (10:36):

It's just that as financial institutions, we've never been willing to let go of that ownership issue because we had no way of answering questions without that ownership issue.

Brian Roemmele (10:48):

And the thing is, we all learn as we get older, you don't really own anything. You're a steward of the things that you have. I've been working on the Wisdom Keeper Project, which is the recording of your first personal wisdom by answering a thousand questions on a voice memo device.

Brian Roemmele (11:07):

I just did a podcast where we revealed this and it's a mission of mine to save wisdom, and to have that wisdom transferred onto the ages. And despite putting on a cheap voice memo device that nobody buys anymore, they're 25, 30 bucks.

Brian Roemmele (11:25):

In your own voice, you are answering these questions that are profound. I've not met a person that does not tear up when they're going through this process. Oh, it's a cathartic process. You're going, "Oh, oh yes." And then what do you do with that?

Brian Roemmele (11:42):

Well, my mission is to build your own Wisdom Keeper, your own personal AI, where somebody, your great-grandchild can come to it and say, "What did great grandpapa Jim feel about this?" And in your own voice, Jim comes talking back. And it's not a bid for immortality, it's not reduplicating you in AI. No, it is a way for somebody to reach across the ages and touch you.

Jim Marous (12:07):

Well, how many of us have wished that we saved the conversations, the thoughts, the humor of a relative that's passed on, and I've actually used it in some of my presentations. The ability sometimes at some point, to at a funeral actually have the person with the generative AI capability of visual aspects actually speak at their own funeral, to share the way they thought about what was going on in the world, what they experienced, stories that you all will remember. But as in the game of telephone, loses something in every iteration.

Jim Marous (12:41):

So, when you look at ChatGPT and generative AI and all the elements regarding AI, we talked about it in part one of this interview, that there's a whole dynamic of it's not really completely converting everything to digital.

Jim Marous (12:58):

So, when you look at trying to strike the balance, how should banks or how do banks strike a balance between leveraging AI for personal experiences, and maintaining the real human touch in a banking relationship? Or are they always going to be completely integrated? Or are they separated to a degree?

Brian Roemmele (13:20):

Wow, it's a tough question, Jim, because I think both kind of at the same time. We're forming the new corporation, the new amalgamation with this technology. And we could see it from the prior podcast, we talked about how this is such a big tool to be utilized in empowering employees.

Brian Roemmele (13:43):

And it's such a big tool for the customer to have access to this big AI model that they're going to have it built for them. Albeit in Cloud, some dangers with that, but again, doing things the right way with transparency, making sure the locks are really clear. When somebody opens that lock, we know it.

Brian Roemmele (14:04):

We know that lock's been open, and then we can ask why. It's like a safety deposit box. So, your AI is a safety deposit box of your financial dealings with that organization.

Brian Roemmele (14:15):

And like I said, you've never really owned that data, you've had access to it. And if you did the wrong thing with it, you overly marketed or you under marketed, you're going to lose that data anyway. Because you're going to lose a customer. The data's almost worthless once you lost a customer.

Brian Roemmele (14:30):

You're not going to acquire them because your brand reputation has been deleted in that person, or diluted to a certain degree.

Brian Roemmele (14:37):

So, what do we do? We go the other way. The other way is what? Let's form a partnership. And I equate it to walking down the street in 1950s America. And you had a local banker and you knew the guy's name, you knew their family, you knew where they went to school. And there was a good feeling, and we need to get back to that.

Brian Roemmele (15:02):

We got away from it because we centralized when we built really big, and we built these cold, distant, structural relationships that served nobody. It was a broken clock that was right twice a day. It's 12 midnight, it's 12 noon, it's working Brian. That's where big bankers went.

Brian Roemmele (15:20):

It's like the ATM machine. The ATM machine was really not designed to replace the teller, but if you get somebody who is just going to sharpen their pencil and look good for two or three or four quarters, yeah, maybe it did.

Brian Roemmele (15:35):

So, why did bankers get into the situation they got in? Because they did not embrace technology the way we're talking about today.

Brian Roemmele (15:43):

Not to throw people into this cold cement and glass structure where they're alone, but to bring them back in, to have a cup of coffee and to hang out, because that is the relationship that is transcending ages, families, generations will go to the same bank.

Brian Roemmele (16:02):

You know, man, heck or high water that say, "No, we're banking with Joe, because we know Joe and he's been with our family, and he's worked with us." The same can be done with AI, but to do it in a careful way. Not to try to bs the person and to thinking, "Oh, the AI is your personal banker and it's going to ..."

Brian Roemmele (16:22):

I'm not talking like that. I'm saying this AI is AI and we, your personal banker, is going to use it to make this relationship that more powerful. And we're going to use this opportunity to open technology so we can get close enough to you that we can be a better service.

Brian Roemmele (16:40):

I think the learning lesson of sticking nothing but ATMs in the lobby was with a death sentence for a bank branch. Because then it became, "Why do I need to go here?" And by centralizing all power and not giving authority to local bankers, there's, "Why do I even need to know my local banker?" I mean, it seems so logical.

Brian Roemmele (17:01):

Now, I tried to explain it when it was happening, but if you're hiring people that came out of Harvard and they are just about making a better two or three quarters and getting a bonus, unfortunately, the brand's going to get diluted. And that's what we're experiencing, not just in banking, but all across corporate America. So, I don't want to go too far on that.

Jim Marous (17:25):

No, but I mean, it's a very good point because what you're talking about, the humanization of this becomes, because of it, becomes relational instead of transactional. Because you're continuously using data to build the relationship, the revenue will come, but the revenue will not be the ultimate goal.

Jim Marous (17:44):

So, on the other side of that, when we talk about the other side of the financial component of efficiency and effectiveness, how can ChatGPT and generative AI be utilized to enhance operational efficiency and to streamline processes?

Brian Roemmele (18:04):

Absolutely, across the board, Jim. I can't think of a process that won't be touched by an internal AI in a way that will empower everybody within the organization to work much more efficiently. Now, what this means from the lowest order is to take away redundant and mechanical type of work out of human hands and to put it into the hands of the machines. This is the history of humanity.

Brian Roemmele (18:31):

You have Ned Ludd who started a ... Later was known as a group of Luddites. Ned feared his job as a weaver. He would put his hands in the weaves, and he would move the shuttle back and forth, and he was mad that his job was going to be replaced by the new mechanical weaves.

Brian Roemmele (18:50):

So, one night when the machines were ready to take off, he didn't know what the managers had in store for him. He went and destroyed all those machines, and they became known as the Luddites. What Ned did not know is that they were planning to put Ned on that machine to train it and to be an operator, so his fingers weren't numb, he wasn't bleeding at the end of the day. We still needed Ned's knowledge.

Brian Roemmele (19:14):

We still needed Ned's Knowledge, he still was materially important for that company. The machine did not replace Ned, the machine made Ned more powerful. So, when we hear the story of the Luddites, and a lot of it is sort of fictional, but I've researched this, some of it is very factual.

Brian Roemmele (19:29):

They were not firing people, they were empowering people. We just see it through the eyes of we have politics, we have unionization, it became a good story to workers being fired. So, things like that. The same can be true within processes within banks.

Brian Roemmele (19:45):

When I sit down with a bank, and I have one banking client I just recently did this with, we looked at every single process, and this is what I would do; you have an audit. What is this process like? Okay, easy. This, this, this is AI - this, this, this is human, and this is a new function for this person, they now have a new position.

Brian Roemmele (20:08):

Now, what happens is, are their jobs being replaced now? Now, they're taking the sum total of all their work, and they're now mentally operating on it and finding new value. And once you cut loose that empowerment, you see an operation become a startup.

Brian Roemmele (20:27):

Literally, an old legacy company can become a startup, because now they have the technical sales force and the technical programming force, and all these other things, marketers. I mean, there's all kinds of things that AI can do that require a human final vote on. But it's producing work, it's producing research.

Brian Roemmele (20:48):

And you just see people, they bounce into work. I mean, this one bank I'm working with, they said employee satisfaction. First off, they were losing a person every month, and this started about nine months ago. And since we've empowered employees, they've not lost a single person. Worker satisfaction went from 50% to 90%.

Brian Roemmele (21:12):

It's hard to get a productivity cloud, it's always going to be nebulous. But their overall metric that they're using for productivity is to the 80 percentile where it was at a 10% productivity.

Brian Roemmele (21:29):

What part is AI? What part is human? I don't care, it doesn't really matter. At the end of the point, if you extract a human from the AI process and you just throw it a bunch of machines, you're probably not going to get anywhere near that capability, and you're probably not going to get the insights.

Brian Roemmele (21:49):

Now, do AI systems make insights? Absolutely. Insights that humans didn't make? Absolutely. But do humans make insights from the AI data that the AI could not possibly conceive of? Absolutely. So, there's a simpatico relationship with the AI.

Brian Roemmele (22:08):

And if financial institutions, banks don't waffle this, and they really take this for maybe the first time in a long time and say, "We have a sacred relationship with our clients, and that relationship can be so much more powerful if we do this the right way. If we take five or six quarters and we take a step back and not just look at what we can sell, but how much better can we make this relationship so that it lasts an entire lifetime."

Jim Marous (22:44):

And beyond generational.

Brian Roemmele (22:44):

That would be the goal, and generational.

Jim Marous (22:48):

So, let's take a short break here and recognize the sponsors of this podcast.

[Music Playing]

Jim Marous (22:51):

So, Brian, what are the limits? I know you've worked with organizations and you're getting deeper and deeper into what the potential is for ChatGPT and generative AI. What are the limitations and potential pitfalls that banks must be aware of when deploying these capabilities?

Brian Roemmele (23:14):

Yes, they're not perfect. They will say things that we don't imagine they should say. There is randomness, there are hallucinations. And this is a reality, and you can wait forever. And there's a lot of companies that are waiting on the edge that because we're a corporation, we can't have this GPT say something wrong.

Brian Roemmele (23:37):

Well, you're going to have to wait a long time. Here's what I say; tell everybody that you are working with modern technology that's safe, but it's also fallible in what it says.

Jim Marous (23:49):

By the way, just like humans.

Brian Roemmele (23:51):

Absolutely. You pick up a phone, and let's just say the customer service is not necessarily in your native language, and you start hearing things that you thought you heard and you don't quite understand, that's happening all the time.

Brian Roemmele (24:04):

And I dealt with one bank that's right now thinking about working with us. And I played them back recordings of their customer service. And they were all from a different part of the world and they were all answering questions that they did not have answers for, was not on the screen in front of them.

Brian Roemmele (24:24):

And I said, if I had an AI system in front of them, they would've answered those questions to a fine level, because here's how I do this with customer service. I'm not of the mind where I want to completely eliminate the human, because I think there is a fallacy in that at this point.

Brian Roemmele (24:44):

I think some people say, "Well, let's just make it a chatbot and be done with it." I think we already proved that we're not going to get there anytime soon. And I think trying to convince people of that for anything of high value, you're probably going to lose brand equity at some point.

Brian Roemmele (25:00):

But what I do have Jim, is we have speech-to-text, as a person is talking, it's presenting a question to the local AI system, so the person doesn't even have to type it in. And once they're stopping, the AI starts producing an output of what it thinks would be the best answer.

Brian Roemmele (25:20):

And that person has a choice to read it back verbatim. If they're capable of more, then they elaborate within the bounds of the nomenclature they're capable of, or they use it as a basis, and it is flawless in its capability working in that way.

Brian Roemmele (25:40):

It will bring up ideas that maybe a person working 10 hours sometimes may have forgotten to bring up. And it may say that, "Bob Smith has called nine times over the course of this year, and he's asked this question a few times, therefore, I'm going to redo how we respond."

Brian Roemmele (26:02):

So, it may say that, it may not, it may just think that up on its own in a sense and represent the idea so that Bob Smith may now understand it.

Brian Roemmele (26:14):

This happens a lot in customer service situations. You get people over a course of days, weeks, and months that are posing a similar or the exact same question. And it could be that they never really fully understood it or they're frustrated by it, or there's memory issues. There's a lot of things that can come from it.

Brian Roemmele (26:32):

But this is giving an immense amount of power, again, if used correctly. Now, if you just made this an AI system and you had no human in the loop and it starts going a little bit off the rails, then yeah, you're probably going to get a little embarrassment.

Brian Roemmele (26:49):

But again, if you're transparent, the transparency is there that your basis is human-to-human interaction aided by machinery, this machine, just like a car is a machine. We're not walking, we're going in an automobile to get from point A to B. This is no different than that.

Brian Roemmele (27:07):

So, if we continue to use that sort of lens to look through it, people are very forgiving. I've been in test cases where we've used AI, and we've told the person that we're going to do this customer service call with the assist of new modern AI, and we're just going to read it back, and we want to gauge if this answered your question. Complete transparency, no games.

Brian Roemmele (27:30):

And what happens is actually, people are quite forgiving, they're very forgiving: "Well, that was great, I wouldn't have thought of that." And we get all those types of responses, but sometimes when it percolates up to the executive level, oh no, we can't do that.

Jim Marous (27:46):

Well, it's interesting too, because when you look at the possibility of ChatGPT and generative AI asking questions to dig deeper, when you look at the masses and combine it with the individual perspectives of dialogue that's going on, you can get in a loop of people asking the same questions or having the same problem.

Jim Marous (28:05):

You go, "There's other people that have had that challenge, maybe I have to ask a question if they've done this." In much the same way, a human does saying, "Are you going through this process this way?" Where in the past that question may not have been asked. It's not simply a generation of outputs, it's the action of inputs as well.

Jim Marous (28:25):

So, Brian, on the last question, this is a great way to culminate two parts of a really great conversation on ChatGPT and generative AI: what do you see ... I can't ask you for long-term future because nobody knows what that's going to be.

Jim Marous (28:42):

But in the short-term, let's say in the next year, what do you see as the most exciting element that may come out of this whole journey we're in with ChatGPT and generative AI? Recently, we just heard that Apple's creating their own version of this, what excites you about the near-term future?

Brian Roemmele (29:01):

Wow, it's so hard to pin this down, but speaking of Apple and all of that, is I think that we are going to get much more wowed by what the capabilities are with this type of technology.

Brian Roemmele (29:16):

I think Elon Musk's xAI is going to shock us in its capabilities because it's built on curiosity. It's built on scientific curiosity of the world, not trying to have any other goal than that. And I think that is what drives all of us romanically, it's our romanic vision of the human future.

Brian Roemmele (29:39):

And I also think that our concept of software and apps are going to change because we're not so much going to have an operating system or an app, we're just going to ask AI to do something. Like "Make a spreadsheet up of this." It'll create a spreadsheet and why would you download an app? What's the point?

Brian Roemmele (30:01):

And word processing, obviously, almost everything you could think of will be self-generated within the AI operating system. The opportunities that that offers - now, we can cling to the past and say, "Oh, my app world, it is like the 299, I could sell an app." It's over. We've reached peak app three years ago. Nobody downloads apps that much anymore, and we know that.

Brian Roemmele (30:27):

So, the future is an opportunity that's grand, and that is understanding what prompts are. Anybody listening to my voice, learn prompting, whether you go through us or ... You don't need anything official, go to GPT-3, do it for free, go to Bard.

Jim Marous (30:44):

Keep playing with it.

Brian Roemmele (30:45):

Keep playing with it and devise your questions and see what you get out of it and notice how it keeps changing. And I would say by the end of this year, we're going to be wowed by a couple of new companies that are going to be bringing out AI that we could not have comprehended, and that's very soon.

Jim Marous (31:04):

Brian, again, it's always amazing to talk to you. I will reveal something as we just had that last conversation. All the questions I asked today were generated from ChatGPT where I gave it the title of what this podcast was going to be.

Jim Marous (31:22):

I said, "Can you give me an intro paragraph? Can you give me an outro paragraph? Can you give me a selection of questions I can use?" We had some adjustments, I had some edits, there's still human interaction. But the reality is it came up with questions that I would not have thought of. It did not have some questions that I thought were good, but they presented it better.

Jim Marous (31:44):

And the reality was, as I've gotten better at asking the questions, the answers have become stunning as Leah and Chris know, because I've shared them with them. And what's great is the future, the potential, the vision of what can happen is really going to still be based more on how humans interact with this tool as opposed to the tool itself.

Jim Marous (32:07):

When I asked a person from IBM about three months ago, "Has the technology gone beyond the technologist that developed it?" He said, it certainly has, it now is in the hands of the public and it's going to get smarter exponentially from the interactions it's having.

Jim Marous (32:24):

Brian, again thank you for going through two different parts of this podcast. It's the very first podcast we've ever had a two-parter, but there was too much to discuss and too much information that you're sharing. Thank you so much for being on the shows.

Brian Roemmele (32:39):

Jim, it's such an honor to be back and look forward to future conversations. Because this is always going to expand and I just absolutely value time with you, thank you so much.

Jim Marous (32:49):

Thank you.

[Music Playing]

Brian Roemmele (32:49):

Thanks for listening to Banking Transformed, the winner of three international awards for podcast excellence. If you enjoyed today's interview, please take some time to give our show a five-star rating. Also, be sure to catch my recent articles on The Financial Brand and check out the research we're doing for the Digital Banking Report.

Jim Marous (33:10):

This has been a production of Evergreen Podcasts. A special thank you to our senior producer, Leah Haslage; audio engineer, Chris Fafalios, and video producer, Will Pritts, I'm your host, Jim Marous.

Jim Marous (33:18):

Until next time, remember; understanding the potential of generative AI is a key to redefining what's possible in the future.