Hosted by top 5 banking and fintech influencer, Jim Marous, Banking Transformed highlights the challenges facing the banking industry. Featuring some of the top minds in business, this podcast explores how financial institutions can prepare for the future of banking.
How Generative AI Will Revolutionize Financial Services
Generative AI is poised to revolutionize the financial services industry. According to new research from IBM, the market for generative AI in finance is projected to reach $9.5 billion globally by 2032. This disruptive technology has the potential to transform customer experiences, empower employees, and rapidly modernize legacy systems.
As we gear up for Sibos 2023, we’re joined by John Duigenan, General Manager, Financial Services and IBM Distinguished Engineer, IBM Technology on the Banking Transformed podcast.
We discuss some of the most promising AI use cases gaining traction, why an open and hybrid platform is critical, how generative AI can aid application modernization, and what financial executives need to think about when implementing this new wave of AI.
Hello and welcome to Banking Transformed, the top podcast in retail banking. I'm your host, Jim Marous, owner and CEO of the Digital Bank Report and co-publisher of The Financial Brand. Generative AI is poised to revolutionize the entire financial services industry. According to new research from IBM, the market for generative AI in finance is projected to reach $9.5 billion globally by 2032. This disruptive technology has the potential to transform customer experiences, empower employees and rapidly modernize legacy systems. As we gear up for Sibos 2023, we're joined by John Duigenan, General Manager of Financial Services, and IBM Distinguished Engineer for IBM Technology, on the Banking Transformed podcast.
(01:11) We discuss some of the most promising AI use cases, gaining traction, why an open and hybrid platform is critical, how generative AI can aid application modernization, and what financial institution executives need to think about when implementing this new wave of AI.
(01:30) As we've mentioned in previous podcasts, there's a huge potential for the use of generative AI in financial services. The key is to find partners who can enable financial institutions to capitalize on these possibilities while proactively mitigating risks. So John, I know we had you on the show before Money 20/20 in Amsterdam, but can you share a bit about yourself and your role at IBM Technology, as well as how IBM works with banks and other financial institutions to integrate generative AI into digital transformation efforts?
Well, Jim, first of all, it's wonderful to see you again. I always enjoy our conversations. My role in IBM is global industry leader for the financial services industry, the general manager, and distinguished engineer of the industry. And our responsibility as a team is to bring to life IBM technology in the context of specific industry needs. You asked the question about how we are bringing generative AI to life in the industry, and I think that's a long discussion, but one that is top of mind for so many of our clients at the moment. Everyone's active and excited about this space, and I think IBM's bringing forward some pretty unique approaches to how we're adding value to our clients and differentiating the solutions that we bring forward.
We're about to go to Sibos in Toronto in a couple of weeks. And there's a lot of talk about AI in every field that we talk about. It's interesting. No conversations can take place without talking about AI or generative AI. As we go toward Toronto in a couple of weeks, what do you see as the state of AI today in the industry? Is it more talk than action, or is there actually a lot of great things going on right now?
That's a great question, Jim. We've done some very specific research to be able to answer that. The typical answer is that everyone's doing it. Everyone's excited about it. But actually we've done some fairly scientific research about this, and we recently did the IBM Institute for Business Value survey called or report called CEO decision-making in the age of AI. Now, this was a cross-industry study that included responses from hundreds of financial services industry CEOs.
(04:04) And what we saw is that financial services industry CEOs are being incredibly selective and deliberate about their use of generative AI. Everyone in the industry recognizes and sees the potential, and over 40% of the 360 banking and financial market leaders that we surveyed said that they expect generative AI, deep learning, and machine learning to deliver compelling financial results over the next three years. The top areas that they're focused on are on talent, security, and, of course, customer experience.
(04:45) They're the areas where we see everyone in financial services being active, and that's where they want to use generative AI. 54% of those industry chiefs highlighted customer care as a top opportunity, and I think that 75% of the CEOs that we spoke to believe that competitive advantage will be seized by the institutions that have the most advanced use of generative AI. So we're fact-based and thoughtful about this. The appetite, Jim, is huge.
So the study looked beyond just financial services. When you looked at the financial services industry compared to other industries, what kind of woke you up a little bit and said, "Oh, that's interesting, the dynamics, the gaps, or the differences between different industries?" We look at generative AI and AI in general.
Several layers to that answer, possibly Jim. The starting point here is that I think financial services CEOs, and CIOs and leaders have been active in AI for some time already. And we talked about this before. Everyone's been experimenting with machine learning, deep learning, natural language processing, virtual assistance. They've been doing that for a while. The learning so far has been that scaling AI has been incredibly difficult, and the risks associated with AI have been complex.
(06:15) We saw some bad decisions being made because of bias models, machine learning models that were trained with bias data in the past, and those became front-page news stories. So there's caution around that. Along comes generative AI, and we just amp up one the excitement and the interest but also increase the need for generative AI to be able to operate in a regulatory context.
(06:45) Now all of the excitement around generative AI came from consumer-grade AI, and that's not necessarily applicable to regulated financial services industries. I don't think hallucinations, wrong answers are going to be acceptable to many chief risk officers or to the customers and clients of those financial institutions. And equally, regulators are going to have so much to say about how generative AI needs to be used responsibly. And so it probably needs all kinds of enterprise capabilities.
So pair it down a little bit. What do you see as the most critical considerations that C-suite and financial services technology executives have to take into account when they're looking to adopt generative AI solutions?
So this speaks to the very heart, Jim, of what we've done with Watsonx, IBM's AI and generative AI offering. And the reason we built that offering is because we've been listening to our clients, we've been listening to senior leaders in the industry, and they've told us that they need solutions like ours to have very specific characteristics. And so the way that we've built Watsonx is to have four main characteristics. First is that AI and generative AI needs to be open. IBM's Watsonx platform is an open platform. That means that we make generative AI a team sport. Not only does IBM create large language models and foundation models, but we also encourage an open ecosystem in which many players have the opportunity to participate. And so you may have seen the news around IBM and Hugging Face. Hugging Face is a provider of generative models, large language models that have specific purposes.
(08:51) Meta's Llama 2 model is part of the Watsonx platform. And so openness is really important for us. Trust. Trust from a regulatory perspective, trust from an answer perspective. And so we believe that a generative platform like Watsonx must have trust. It cannot have hate speech, it cannot have access to information to which it's not entitled, for example, copyrighted material. And in the course of trust, we need to be able to ensure that answers are correct and explain how an answer was generated. That's the essence of trust and bias-free and inclusive.
(09:38) We are firmly of the belief that a platform like Watsonx must be targeted. Targeted for very specific enterprise use cases. So not the be all and end all of everything that writes poems and music and draws pictures, but a platform that creates value by solving actual, important crucial business problems. And last but not least, the fourth characteristic of differentiated enterprise AI is that it empowers people. It's empowering. Empowering in that what we do with AI uplifts society, uplifts people's job roles. And so when we think about generative AI and how we bring that to the enterprise, it really is around a platform with capabilities that are open, trusted, targeted, and empowering.
You know, it's interesting, John. We talked just a few months ago, and we were just talking about the very beginning of the integration of Watsonx within a generative AI framework. The industry's going that fast no matter who you're talking to.
Do you see as financial institutions are out there trying to solve numerous problems out there that the challenges they have in digital transformation? Do you see them looking for one organization partner for the degenerative AI solutions? Or do you see the possibility that they'll pick and choose solution providers based on what they're trying to solve, for despite the power of one solution versus another? Because, like a horse race, these things are going to jockey for position quite a bit over the next few years.
Yes, I'm sure they will. We are very clear that we are laying the table with IBM's value proposition, and we recognize that clients have choice and clients will want to experiment and learn across a whole range of opportunities. Generative AI will be built into all kinds of products and capabilities that our financial services clients will provide, but the essence for us is that by providing that open platform, we're providing choice. And the other thing that's crucially important to mention here, Jim, is that because Watsonx is part of IBM's hybrid approach, we are also giving choices around where the AI workloads get done. Many of the technology providers in the generative AI space are forcing clients to use a specific cloud service or a specific software as a service. That's kind of okay, but that implies co-mingling your data with others potentially. And co-mingling your learnings and insights.
(12:37) Also, they might be your proprietary secret source, so why might you want to do that? The thing that we've done with Watsonx is ensure that it fits in with our hybrid way of delivering software, which means that instead of bringing the data and the models to the AI, we are bringing the AI to where the data sits and runs today inside the customer's environment, in their data center on their premises. That's unique in terms of generative AI offerings. And so, as much as I love that clients will have choice, IBM set the table in a very specific way that lines up with our core mission of being the hybrid and AI company for the enterprise.
So as you look at what's going on right now, I mean there's obviously some use cases, some integrations that are starting to take shape. You are at the forefront of what's going on here. It is very obvious. What excites you the most about what financial institutions are doing now? And just as importantly, what they'll be able to do in the near future?
Well, I've been all over this space. The first use case, we are looking at this from the point of view of three specific use cases, Jim. The first is customer care, second, digital labor and talent, and third, really around application platform code modernization and engineer productivity. And so when we think about customer care, that's the starting point. As I said up top, that's where everyone's starting point is. Now everyone has virtual assistants today, those virtual assistants have been put together in quite a complex way. The power of adding generative AI to a virtual assistant is that we can expand the range of questions that a virtual assistant can accurately answer without a massive training process. So what we've done with a number of clients already is point Watsonx at a trusted repository of information to train a model and then directly connect the virtual assistant, Watson Assistant, to that Watsonx capability.
(15:00) And have it answer questions based upon a knowledge base that is maybe sourced from a financial services firm's website, where they outline their policies and procedures or product capabilities. With very minimal training, we can add way more value to a virtual assistant than the intents that have been pre-programmed into it. So just as a starting point, that becomes very, very interesting. As we move on into the realm of digital labor, that's really about how we optimize and orchestrate business processes.
(15:39) That space is ridiculously exciting for us simply because of the potential to take friction out of business processes, take human handoffs that often go wrong out of business processes. And so that's a huge area and modernizing platforms. Everyone wants to do that in support of their digital transformation. So we think about those three specific use cases. It's not that there aren't a thousand more, and there will be, but we all need to start somewhere. So we've started on the top three, where we see firms deriving the most value from the Watsonx platform.
So let's take a short break here and recognize the sponsor of this podcast. Welcome back to Banking Transformed. So I've got to ask you a question that's maybe not banking-related, but you just wrote about it in LinkedIn, an experience you had in trying to buy a very nice TV, and it prompted you, as interestingly, those human things that happened to us prompted us to think it's got to get better than this. But you talked about your experience and how there were so many broken experience promises based on what you knew is possible. Tell us a little bit about that case study of yours, and then a little bit about how you think generative AI could have made it better.
Well, it's so much fun that you're asking that question, Jim. So I did buy a lovely TV. It is lovely, I'm so delighted with it. But the fulfillment process was an absolute nightmare. And I already wrote in the LinkedIn article why, and why there were so many steps to making it right. The sales process was fabulous. It was easy. It was four clicks, and I was ready, and I customized it the way I needed. Unfortunately, from the very next step on, things went wrong, the delivery date changed. The delivery people didn't know what they were supposed to do. And then all of my interactions with customer service after that were horrendous. I needed to authenticate multiple times. I needed to explain the situation every single time. And then customer service agents were not empowered to make the answer. All these same things that I experienced buying and delivering a TV can happen in financial services every day.
(18:10) And this is really about disconnected information, disconnected processes, and unempowered people. Generative data platforms like a data fabric and an AI platform can integrate information effortlessly. That's what they do, right? They solve that fundamental problem of someone looking at the screen, not knowing what happened, not knowing the status of something, not knowing the history. And then, of course, if you can orchestrate the business process, you remove their human handoffs where someone needs to call someone and say, "Are you delivering this product? The customer's on the phone with me right now, and they're unhappy. What are you going to do about it?" Right?
(18:52) All those internal processes that I should never know are happening become very clear to my eyes. And so the ability for a generative AI platform and business automation to orchestrate those processes and integrate the information, and make status transparent and obvious to a consumer or a client, that's where the power of these technology integrations come together. And so when you think about generative AI and orchestrated business processes, that's a dream, and that's how to deliver a really perfect fluid, frictionless customer experience that we all crave.
(19:34) Given that customer experiences top-of-mind in customer services. That's exactly why I wrote about that broken experience, because everything I could say about that also applies to how each one of us interacts with our banks and insurance companies.
You know what's interesting, John, is just because you're close to what's possible doesn't mean that all consumers aren't close. It's interesting, you should talk about the article you wrote for LinkedIn. I wrote an article about an experience I had with Delta Air Lines, who usually is extraordinary with their customer care, and my end solution was extraordinarily good, but getting there was so painful.
What's going to be interesting is every consumer, as they get more and more aware of what generative AI can do and what OpenAI is doing. ChatGPT is doing in a consumer world, their expectations are going to escalate so fast that their-
- You touched base on a really interesting component that I don't think we take enough time about is that that generative AI can really change not only the productivity of the employees in the back office but the overall employee experience. Can you touch base on that just a little bit on what you see as the vision for the future there?
Fabulous question, Jim. Yes. We obsess about customer experience all the time, so much so that we often don't think about employee experience. You're right in saying that all of us have been trained on contemporary customer experiences just with the vendors that we interact with pretty much every day. Depending on what supermarket you go to, whether you use eCommerce websites, your phone provider, every one of those firms have tried to take friction out of the process and get to know their clients with relevant insights. Oftentimes, all the investment has gone into that experience. But employees behind the scenes who are also trained in all the same customer expectations don't benefit from that same technology on the inside. So think about a customer service professional if they don't have ready access to information that helps resolve a customer's problem, helps provide a solution, helps provide status.
(22:14) Not only do they get frustrated, they're not in the best position to serve the customer or the person who's calling in asking for help. Think about all of the manual and menial work that goes into all kinds of business processes that employees undertake every day. There's some really great examples of this. HR is full of those examples of where there's a massive amount of information to be integrated in making hiring decisions. All sorts of investigative processes inside a firm are also subject to that information integration and automation. Whether it's the know-your-customer processes of a bank, whether it's investigations and dispute resolution, it could be so many points of contact where an employee or a worker has to touch those systems. And so this is the point about empowering. We are trying to uplift those workers and add higher value to the function, and the work they do, the service they provide.
It's interesting you mentioned how this can help alleviate the need for employees that do certain things today that is rudimentary. I mean, it can actually make them more productive, make them happier at their jobs because they're going to have more important roles. But as we know the workforce right now, the employment situation, we usually are happy about low unemployment rates, but in the talent pool, it gets really difficult to find the right people to do the jobs that are involved with generative AI and AI in general.
(23:58) When you look at that, do you see this as an area where organizations of all sizes are going to have to partner with third-party solution providers that are really capturing a lot of this talent around these new technologies? Because it's going to be hard, I think, for most finance institutions at almost any level to get the best talent to work in banking as opposed to with the technology. Even though they're, it's different sides of the same coin, so do you see the future really being where, in this type of solution, organizations are really going to have to partner with those organizations that can get the best talent to deliver the solutions that you're talking about with not only generative AI but Watsonx?
Yeah, this is going to come down to ecosystems and part of the reason why an open platform is so important and models that are open, and trusted and targeted are so important is because they facilitate building that ecosystem. There will be some institutions that want to build their own models and want to do that in a very specific and closed way where they don't need outside help. Others will say, "I want these three banking industry-specific models from IBM, I want to incorporate that with other models." And they may want to derive their own models from models that we provide. An open ecosystem where AI is a team sport, not just something that we do in the corner, but a very highly productive team sport that's focused on business values and business outcomes is what will alleviate the talent rush that is sure to come here.
(25:47) And we are very excited about that ecosystem approach. It's very fundamental to how we at IBM go to market. And so that's exciting for us building that ecosystem, saying this is more than what we alone will do by ourselves. This is about what we will do with Watsonx together. And that's why all of the firms in our ecosystem not just they write software platforms, they deliver capabilities to their clients. They're also thinking, "How do we use Watsonx for the future of customer service solutions that they build or onboarding solutions?" And so the ecosystem is where this does all come together. It's not about what happens in one-off situations, but it's about how we all build scale together and deliver value, most importantly.
So, John, usually in these podcasts, I in the past, you would ask people. So what do you see happening in the next three to five years? The reality is when we're talking about this subject matter, three to five years, we have no clue. We know what could be done, but really, what we're more interested in is what are we going to see in the next 12 to 18 months? So what is your vision of where we're going to be 12 to 18 months from now? And just as importantly as our final question, what should financial institutions do today to take that step forward?
Well, call your IBM representative, obviously, Jim. Let's talk about that 12- month, 18-month horizon. The most exciting thing that I'm looking at alongside those things that we talked about with customer care and digital labor. I don't know if the viewers saw our announcements about the IBM Code Assistant for Z or Zed, the mainframe system. I am so excited about the potential of the code assistant for Z. We have a set of capabilities where we can now analyze and understand the structure of some of the vertically integrated, hyper-integrated monolithic code and start to find ways to modernize that COBOL code, either into better COBOL or even well-formed Java. There's been a ton of work on this in the past that has led to some situations and approaches that some in the industry would call JOBOL, Java that looks just like the COBOL statement, that it replaced. JOBOL and COBOL, JOBOL especially is not very maintainable, so it's not really a sustainable approach.
(28:38) The kind of COBOL and Java that we are creating with the Watson Code Assistant for Z is well-formed Java. And IBM research have done some amazing work around accuracy. And I will just say that the COBOL model that IBM has built, which covers many programming languages actually, is one of the most accurate anywhere. So if I think about the potential and the possibility of opening up some of that very monolithic code, that fills me with excitement because that's a direct on-ramp, Jim. And you know how important this is to the transformation and modernization that we all want to do. So clients should definitely see that and understand what the impact of that capability provides them to extend the value that they have from platforms like the IBM Z system and their middleware and software that runs on it. In terms of practical steps, I was somewhat joking when I said, "Call your IBM representative." The real practical steps that we'd encourage clients to engage in, to engage in the conversation with us to co-create with us.
(30:01) I love to mention that IBM client engineering is a fabulous way for our clients to experiment and co-create with us. It's a short way for them to gain meaningful experience with IBM technology capabilities in the context of their specific business situations. And so working with IBM client engineering is a fabulous way to explore and understand the capabilities of Watsonx from IBM. And create a meaningful business outcome, a minimum viable product that could be deployed very, very quickly. So that's how we want people to get started, Jim.
It's so exciting. I love banking, but I also like some of the technology that's coming out right now, and the integration of both makes for a very exciting time. On a personal note, I am really looking forward to finally meeting you in person in Toronto at Sibos in a couple of weeks.
Yes. And there's nothing to be said about your lack of passion. You have no lack of passion with regard to what's happening. And what's interesting to me is, every day you open the journals, or the research papers, or the news in general, and you see so much happening in the industry, and you can envision where this can go. And we are going to get you on again in the near future because continually being updated on what the potential of generative AI is, what the potential of AI is. And what the potential of the integration of all the technologies that are out there that the financial services industry can use to improve the way consumers manage their money, do their banking, and basically improve the wellness in a way that's never been thought possible on a personalized basis. The future is really bright.
(31:55) John, thank you again for being on the podcast. It's always a joy talking to you, and as I said, I'm really looking forward to seeing you in Sibos, in Toronto, in a couple of weeks.
Thanks for listening to Banking Transformed. The winner of three international awards for podcast excellence. If you enjoy what we're doing, please take 30 to 45 seconds to show some love in the formal review. It helps us to continue to get great guests like today. Finally, be sure to catch my recent articles in The Financial Brand and check out the research you're doing for the digital banking report. 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.
(32:42) I'm your host, Jim Marous. Until next time, remember, AI has the power to unlock incredible capabilities that can reimagine the future of banking.