Subscribe to “The Banking & Payments Show” podcast on Apple Podcasts, Spotify, YouTube, Podbean or wherever you listen to podcasts.
Episode Transcript:
Jacob Bourne:
It's one thing to have a leading AI model, it's another thing to be able to integrate it into a chatbot that is going to be performed well for a specific use case and for a specific demographic. Customers who might have different language skills, for example, come from different backgrounds, be at a different financial place and their history. So it's just about having that user experience be tailored to whatever customer it is and make sure they have a good experience using it.
Rob Rubin:
Hello everyone and welcome to The Banking & Payment Show, a Behind the Numbers podcast from eMarketer. Today is February 6th, 2024. I'm Rob Rubin, head of business development at eMarketer and your host today. Today we're going to discuss what to expect in terms of deployment of AI in banking this year in 2024. And for this discussion, I invited two of our analysts at eMarketer who have been focusing on AI, Gadjo Sevilla and Jacob Bourne. Hey guys, how you doing?
Gadjo Sevilla:
Good. Happy to be here.
Jacob Bourne:
Glad to be here. Thank you for having me.
Rob Rubin:
Yeah, I've been trying to get you guys on the show for a while, so I'm really happy to have you. Because we've never had you on the show, I want to break the ice with a few questions about your experiences with ChatGPT. First, what's the weirdest thing you've experienced?
Gadjo Sevilla:
I can start with that. When it first came out, I asked ChatGPT who I was, so I typed in my name and it completely hallucinated.
Rob Rubin:
Who were you?
Gadjo Sevilla:
So I was a real estate broker based in Toronto, Canada, which I did live there for a while, so that part was correct, but it said that I had gone to University of Toronto, which I didn't.
Rob Rubin:
It's a good school, by the way.
Gadjo Sevilla:
Great school. Yes. It also said I had three children, which was completely incorrect. So that kind of threw me for a whirl. First time experience.
Jacob Bourne:
I wonder what it would say now. I think that a lot of the really weirdest experiences come from trying to intentionally jailbreak the thing, which I haven't spent a lot of time doing that, but a couple times I've had it open in my browser with no prompt, no conversation going, and all of a sudden it'll start answering a question I haven't asked, which is always a weird freaky feeling. And it's like a question that I probably wouldn't ask either. And so I've asked it too, I'm like, well, I didn't prompt you. What are you doing? And it will apologize and say it's some error in the system or something. I always wonder if somebody else had maybe prompted that question. But yeah, I have used it in my personal life as well as at work. I do find that if you want a quick conversational answer to something that's going to take you a while to do it via an internet search, then it's definitely a go-to method to get that kind of information.
Rob Rubin:
Great. I want to jump right into our topic today. Instead of talking about recent headlines in the news, let's chat about some of the use cases that we're going to see maybe in 2024 in banking. And there's sort of five that I always read about and hear about, customer service and chatbots always, personalized banking services, being able to help your own customers figure out what their banking services should be, what they should do, reminding them of things. Obviously fraud detection and prevention is one that comes up all the time, as does credit scoring and risk assessment. And then the other is sort of personalized marketing, which I personally think is going to get a little creepy, but those are the sort of five top of mind ones. Do you guys know, are any of these use cases already out there that have been deployed?
Gadjo Sevilla:
Not sure if they're already out there, but MasterCard just announced that they're jumping into generative AI and they're hoping it can boost fraud detection by up to 300%. So I mean, they're going all in on that and I'm expecting other banking institutions to follow suit as well.
Rob Rubin:
So you think the fraud part, what about the chatbot, which we hear about all time?
Jacob Bourne:
I would say all of these use cases exist to a certain degree in the banking industry. The distinction here is, is it AI or is it generative AI? AI has been used by these institutions actually for years now without it being a topic of conversation. It's kind of upgrade, if you want to call it that, it's kind of debatable, to generative AI that's really in question. And we have seen that with customer service chatbots, you're seeing more conversational responses, personalized banking services, again using that kind of predictive AI, course Gadjo mentioned fraud detection and prevention.
Rob Rubin:
Yeah.
Jacob Bourne:
So all of this, the personalized marketing especially, that's generative AI. That's something we're seeing more of across various sectors, including banking. I think it's also important to note that banking and just the financial industry in general is one of the sectors that is actually seeing a lot higher adoption of generative AI than other sectors.
Rob Rubin:
So what skills does a bank need to execute these use cases because they're different?
Jacob Bourne:
Right. Yeah. It is a lot of different things. I mean, for the fraud detection, that's again looking at preexisting AI. And so kind of the more data science field that's been around for years, the more conversational generative AI uses. You're really looking at hiring AI software developers who know how to build AI models. And I say that because these types of institutions aren't really going to want to, I mean, to a certain extent they might, but they're not going to want to really rely on off the shelf solutions like ChatGPT. In fact, we've seen that some of these institutions have actually banned employees from using it. So what that means is they're going to really need to rely on hiring highly skilled AI talent.
Rob Rubin:
So they're going to build their own LLMs.
Jacob Bourne:
Build their own or use an open source and build off of that. I would really though think that the leading institutions, like if you're looking at a financial institution like Goldman Sachs for example, they're building some of this stuff in-house for sure. And so that really requires leading AI talent and they're competing with big tech, with AI startups to attract that kind of talent. But yeah, and beyond that, it's also looking at user experience because it's one thing to have a leading AI model, it's another thing to be able to integrate it into a chatbot that is going to perform well for a specific use case and for a specific demographic. Customers who might have different language skills, for example, come from different backgrounds, be at a different financial place and their history. So it's just about having that user experience be tailored to whatever customer it is and make sure they have a good experience using it.
Gadjo Sevilla:
Yeah, the advantage the banks already have are the data sets that are required to train their AI, whether internal or external. So in the case of MasterCard, which I mentioned earlier, they have their proprietary algorithm and they're training that on data from what they're seeing is 125 billion transactions that go through the company's card network. So in terms of raw data, that's something they've already captured. It's just a matter of how they can apply that to their specific needs.
Jacob Bourne:
And I would say the one other thing is just when we're talking about financial institutions, they really need to focus on cybersecurity and data privacy even more than other sectors. And so that's really being up to speed on GDPR and California's digital privacy laws if they're operating in those areas as well.
Rob Rubin:
Do you think that all the hype, and there's been a lot of hype right now around AI, do you think that it impacts consumers, what they expect from you, especially if you publicly say, we're using GenAI to do this? So how does that sort of impact their expectations and how should a bank manage that?
Gadjo Sevilla:
I think the consensus seems to be that banks and credit unions will likely be careful about adopting AI solutions, and of course marketing the fact that they're adopting AI solutions. So they could focus on assistive AI technologies, ramp up on machine learning and model building for future products instead of generative AI, which to a lot of people is still unpredictable. The results that can come out of generative AI, they could be way off.
Rob Rubin:
I don't think consumers basically know the difference unless you're in the tech space.
Jacob Bourne:
Yeah. And it could be for some consumers, the hype will translate into a situation where they're expecting something that the AI isn't going to deliver. And I think that that's something that banks have to be wary about. And I think there is pretty bullish adoption among banks for this technology. And so I think what it comes down to is a need for gradual implementation. In other words, you don't need to integrate generative AI into every aspect of your banking business overnight. Slowly introduce it to customers, I think is the way to go. Maybe you do a test with a subset,-
Rob Rubin:
Right.
Jacob Bourne:
To see the kind of reaction before you just roll it out and then have it be, again, something that doesn't live up to consumer expectations. I think it's also about giving people the opportunity to give feedback to a human.
Rob Rubin:
How will they know if it's a human?
Jacob Bourne:
I mean, that's the thing, right? And so generative AI is getting to that point where you can't distinguish. But I think it's also just being transparent, and this goes back to this idea of the, sort of the legal risks for banks.
Rob Rubin:
Yeah.
Jacob Bourne:
Having a legal team that is aware of AI and can deliver that kind of transparent communication that's going to be needed if something does go wrong, to try and reduce the risk of legal fallout from this.
Rob Rubin:
Yeah. Hey guys, now let's play a game. Place your bets. Given everything that we discussed, I'm going to make four predictions about what could happen in 2024, and I'm giving you each a total of 10 points to distribute between the four predictions. And I want us to rank the relative likeliness you think each prediction will come true. So let me read the predictions and then I'll start with you, Jacob, and then we'll talk about what points would you give to each one and the likeliness that it will happen. So the first one is that news stories about over zealous chatbots puts a chill on banks rolling them out. So that's my first.
The second is that bank regulators squash attempts to use AI for investment advice. The third is that deployment of AI enables banks to initiate massive internal layoffs. And the fourth is that small banks and credit unions are able to win more customers because of their deployment of AI for customer service. So let's start with the first one. What would you say? Are we going to see a story about some crazy overzealous chatbot and it's going to put a chill on banks rolling them out? Give me a points.
Jacob Bourne:
I gave three points to this. I think there is a risk there, but I don't think that there's any bank that's going to roll out a chatbot without pretty stiff safeguards in place. So I don't think we're going to see a really sensational story come out at a bank.
Rob Rubin:
So you give it,-
Jacob Bourne:
Yeah. So a three. There's going to be some nervousness around deploying it, but I don't think this is going to stop them.
Rob Rubin:
Gadjo, where are you at on this one?
Gadjo Sevilla:
I'd say a two. For one thing, I know a lot of people don't like interacting with chatbots. The first thing they ask is, can I speak to a representative or human?
Rob Rubin:
Right. Right.
Gadjo Sevilla:
So with that knowledge, I don't think they'll put a lot of investment in that.
Rob Rubin:
Yeah, I'm going to go two as well. So we have Jacob at a three, and you and I are at a two. The next one, bank regulators squash attempts to use AI for investment advice.
Jacob Bourne:
I gave that a two. We will see some regulation around use of AI, generative AI. In the financial sector, I don't think that they're going to outright ban the use of the technology for investment advice. I don't see that happening.
Gadjo Sevilla:
Yes. Same with me. I give that a three. I think it's early days for that unless something goes horribly wrong where bank regulators feel they have to step in.
Rob Rubin:
The thing that concerns me is that they give people advice that doesn't necessarily correlate to who they are or they give advice that sort of falls along racial lines potentially, and that becomes a problem. So for me, I'm going to give that one a three. Then deployment of AI enables banks to initiate massive layoffs.
Jacob Bourne:
This is my highest score. I gave it a four. I think this is highly likely, and actually a report just came out from The Burning Glass Institute.
Rob Rubin:
I read that too in the Washington Post today.
Jacob Bourne:
Biggest impact on white collar workers in banking and tech. And the thing is in tech, we're already seeing those layoffs. It maybe hasn't happened, quite tied into AI so overtly, but there's a huge risk for workforce disruption, I think in banking due to AI. And again, it comes down to the cost savings, payrolls [inaudible 00:11:54] companies spend go to their payrolls.
Rob Rubin:
Yeah.
Jacob Bourne:
So it's a huge area that they can just cut costs.
Gadjo Sevilla:
Yeah, this is a four for me as well. I mean, I think unfortunately this is a trend we're seeing not just with banks but across various sectors. And I think a lot of companies will tend to lean into sort of replacing skilled workers with workflows.
Rob Rubin:
I'm giving it a five. I think it's just going to happen. I think they're waiting for it. We're going to see it. And I think that the younger professionals, people coming out of college where they were using those folks to do a lot of the work is going to get replaced, and it's going to be harder for college graduates to get those sort of entry level positions within the financial services industry. So I'm going five there, sadly. And then the last one is that small banks and credit unions are able to win more customers because of their deployment AI.
Jacob Bourne:
Yeah, I gave this a one. I think it's going to become a necessity for them to do that. But sort of the larger competitors are going to be doing the same thing, and so they're going to need to deploy it in a way that makes them stand out, which is difficult to do again, because the cost of the talent required and the knowledge base required to effectively deploy generative AI is high. So is there a room for some customer gain? Yes, but I don't necessarily see it making a huge difference.
Rob Rubin:
Yeah.
Gadjo Sevilla:
Yeah. One for me as well. I see these banks and credit unions as being rather conservative when it comes to new technology. So I think they will approach cautiously. I mean, it's not saying that they won't adopt it at some point, but I think they're likely be a few years down the line once they see there's some demonstrable effect to that investment.
Rob Rubin:
Yeah, I give it a one as well. So let's just sort of review the answers then to come to a consensus. If we look at the one that we think is most likely to happen, that deployment of AI enables banks to initiate massive layoffs. I gave it a five and you guys gave it a four. So we're at 13 points out of a total of 30 that we could have distributed. The other two that are sort of the next likely is the news stories about overzealous chatbots, putting a chill. You guys gave them a two. I gave it a three. So we're at seven there. The other one, which is a little different is that bank regulators' is going to squash attempts. I gave that a three and you guys gave it a two. Together it adds up as a consensus to a seven.
And then the one that none of us think is going to happen really is that small banks and credit unions are going to be able to win more customers because of their deployment of AI for customer service. And that was at a four. Guys, this was super fun talking about this today, and I want to thank you guys, and I want to thank everyone for listening to The Banking & Payment Show, eMarketer podcast. Also, thank you to our editor, Todd. Our next episode's going to be on February 20th, so be sure to check it out. See you then. Bye guys. I'm going to have you on soon again.
Jacob Bourne:
Thanks a lot.
Gadjo Sevilla:
Thank you.
Rob Rubin:
Awesome. Thank you.
%20scale(0.100000,-0.100000)%22%0A%20%20fill=%22%23000000%22%20stroke=%22none%22%3E%0A%20%20%3C/g%3E%0A%3C/svg%3E%0A)
