We recently caught up with George Neal, PrecisionLender's EVP of Analytics, after he returned from the BankAI Conference, put on by American Banker.
In this week's Purposeful Banker podcast, George takes us through some of the hot topics at the conference, including:
- Why were bankers talking about Iron Man suits?
- Is AI in banking more than a retail story?
- Will automation actually create jobs?
Helpful Links
Build Ironman Suits, Not Terminators - Carl Ryden at Business of Software Europe.
BankAI Conference - Website for the American Banker conference.
Buy vs. Build - Making the case for outsourcing critical bank technology needs.
Podcast Transcription
Jim Young: Hi, and welcome to The Purposeful Banker, the podcast brought to you by Precision Lender where we discuss the big topics on the minds of today's best bankers. I'm your host, Jim Young, Director of Communications for Precision Lender. And I'm joined again today by George Neal, our EVP of Analytics.
George is back in the offices this week after speaking, along with Tim Shanahan, at the BankAI Conference in Austin, Texas. BankAI is put on by American Banker and the event title is, well, pretty self-explanatory. So what were the big takeaways from the conference? Well, that's what George is here to provide today.
George, let's just start off with a general question. Your first time at the BankAI Conference, what were your impressions? And was it what you expected?
George Neal: My impressions were overall very positive. I think that the American Banker team put on a fine conference. It was informative and the speakers were overall quite good. I was kind of expecting the conference to be a little bit larger, given the nature of the topic and the prevalence of AI in the banking conversation these days. While there was a great turnout, genuinely I was expecting the conference to be a little bit bigger.
Jim Young: So I was actually just looking at Carl Ryden's talk at Business of Software Europe, and I will make sure to link to that in our show notes. And he speaks about a concept of building Iron Man suits, not Terminators, when it comes to AI. First, can you explain that concept? And then second, did that concept have receptive ears at BankAI?
George Neal: Certainly. So the concept itself is relatively simple. If you consider an Iron Man suit, what you're talking about is an AI, and I think in the movie its name is Jarvis, encapsulated in a set of technologies that help Tony Stark, Iron Man, perform. They don't replace Tony.
They do not step up and make the moral judgments, and they don't establish the empathy.
They don't make human connections like Tony does.
By contrast, consider the Terminator, it is proposed as a replacement for humanity. The goal of it is to displace the human function.
And so you have this contrast of AI used to augment and support human interaction versus AI meant to replace and compete with humans.
That topic did come up a few times in the BankAI conference. And the general consensus agrees with Carl. The goal is not to replace, it's to augment, improve, and to magnify the human elements and the things which humans do distinctly well. Again, the general consensus comes back, you build AI so that computers do the things that computers do frankly better than humans. They're very good at repetitive tasks. They're very good at basic math, even advanced math. They're very good at simple pattern matching. And you have humans do the things that humans do the best.
Some of the conversations sat around how do you use AI in a customer service mode. And they talk about using sentiment analysis AIs to give the customer service rep more feedback on how their influencing the mood even of the customer calling in.
I thought that was very powerful. If you think about it, having a notice ahead of time, "Hey, based on the way this customer came in, they're quite upset even thought they're trying to cover that up," and being able to assess as a customer service rep, "Hey, I took someone who was quite angry. And at the end of our conversation, they were overall fairly satisfied." Isn't that the goal of your customer service rep?
Jim Young: So are they doing that, just this specific instance, are they doing that sort of thing by word choice? Is it volume if they're recording it? How does that work?
George Neal: So they went into only a little bit of detail on that. But if you look at the space, what they're going to do it is by word choice, by inflection, by speed of conversation, all of those things come into play. And if you think about it as a person talking to another person over a phone, because that's what you have, you don't have facial expression and body language to go with. But when you're listening to someone to assess whether or not they're happy, you listen to how they end their words. Do they end in an up note versus a down note? Are they terse in how they speak? The word choices that they use of course come into play. But hugely, it's about how they say the things.
If you really consider, "I want to talk about my checking account," is a different thing than, "I need to talk about my checking account." Those words are not that different, but how they're delivered certainly are.
Jim Young: Yeah, it's a good point. That sounds like technology for the retail banking side. I guess one question - that I'll probably have several times - is about applications to the commercial side. Or, to what extent at this conference were the conversations about retail technology and to what extent, when you were listening in or in conversations, did you see connections to the commercial and business side of banking?
George Neal: I think a lot of the human interaction points that were talked about were retail-driven. There was a lot of focus on how to manage branch interactions, the absence of branch interactions, and other channels with consumers. But a lot of that technology is still applicable I think to the future of commercial banking.
There was also a focus on operational discipline and the application of AI to operations and operational excellence. That certainly applies to commercial banking where a lot of the friction is found once the initial term sheet is generated. And so that whole process through underwriting and making certain that the final terms match what the customer and the bank are both looking for. A lot of AI can be applied in that space.
I also think that when you talk about something even like the sentiment analysis we were just discussing, I can envision a future where your phone is listening to the conversation you're having with your client as a commercial banker and guiding you to which deal points they're genuinely most attached to. Perhaps you're sitting there and your phone is tapping you on the shoulder, the equivalent of that, saying, "Hey, they were talking about rate, and they were talking about duration. But when he spoke about payment, that's when you should really pay attention."
Jim Young: Interesting.
George Neal: I can envision that as a future not too distant off.
Jim Young: So the banks that were there, obviously a lot of fintech there. The banks that were there, were their conversation around this technology is something that we are building or we've got our innovation labs and we're doing this? Or was it more about were they looking more for partnerships?
George Neal: There were a little bit of, "Here's something cool we've done ourselves." And most of the banks that presented the, "Hey, here's something cool we did ourselves," it was more along the lines of truly experimental thinking. We wanted to explore some completely arbitrary potential use of AI and I would say about 50% of those were successful. You heard those conversations from very large banks that have the luxury of putting together effectively a skunkworks team.
Most however, I would say 95% of the speakers and participants, came to the agreement that it is some level of partnership that is going to make this successful. The agile capabilities and the abilities to react and implement that are present in fintech organizations combined with the capital structures and the capital advantages that are present in a bank and the established presence and relationships of banks, that synergy is where all of that is going to happen. And as much as I don't like that word, it genuinely applies in this case.
So what we heard from most banks is a general understanding of they're not going to be able to do it themselves across all the channels and all the operations that they need to, so they're looking for subject matter fintechs. So again, you go back to a fintech specifically targeted at sentiment analysis for relationship management, a fintech specifically targeted at next best product in the retail space, a fintech specifically targeted, as we are, at commercial loan pricing. So do you see that they're seeking out subject matter experts at the problems they're trying to tackle.
Jim Young: So you mentioned a little bit of the Terminator stuff. A lot of times when people think of AI, the fear is that while automating tasks, there's a fear that that eventually leads to essentially taking your job, being replaced by a robot. We've all heard that sort of thing. But you heard some talk at BankAI about jobs that AI might be creating in banking. Can you elaborate on that?
George Neal: I can. And I heard two things that I thought were really exciting in BankAI. We've had a lot of conversation in this space about the jobs that are going to be created. But to point at a job and say that job was created by the technology that people are worried about, I saw a great example of that.
One of the banks presented and made the comment that they're actively putting together the job description for a bot manager. So an individual whose sole job it is to keep their bots relevant, returning the right return on investment that they're supposed to, as the processes that the bot supports change making certain that that bot is adapted to and ready for that change.
Jim Young: Wow. We can do a whole podcast on how you pull that guarantee off.
George Neal: But if you think about what you're encouraging in that. If in fact your job can be automated, then perhaps you're the best person to do that. And if you succeed in automating your job away, you've done your company a service at your own expense. Shouldn't you have an insurance policy for that? And I thought that policy was very powerful.
Jim Young: Yeah. That absolutely is. Again, love to see how they pull that off. But it's a great concept, definitely. Micro AI, also was a hot topic, what is that? And what are some use cases?
George Neal: So one of the topics that came up, and I think I used the phrase micro AI with you when we were discussing this podcast initially, but it's very small AIs aimed at very specific problems.
So if you think about an anti-money laundering AI, most people think of that as a subject matter expert. But break that down into an AI that specifically looks for deposit structuring, then an AI that specifically looks at transfers, an AI that specifically looks at relationship aggregations. So these are micro AIs within that space, but it allows them to specialize in a very specific and narrow area and bring tremendous value to it.
Jim Young: All right. So this is all really interesting stuff. But let's put me in the role of a commercial banker who is honestly just not that interested in the how here; AI, machine learning, whatever. Can you tell me as a commercial banker, what mattered about this conference to me?
George Neal: Certainly. I think that you should expect, as a commercial banker, there's a lot of focus being placed on streamlining processes with AI and making certain that the time for your delivering of a loan or a deposit relationship to your client should be going down. If your organization isn't doing that, you should be looking to figure out where they are applying those resources because certainly there's a lot of effort going into that space.
Similarly, you're going to see a lot of, I believe, improvement in credit estimation and risk rating estimations. There's a lot of AI and technology being put around identifying credit risk up front. I think that similarly you're going to find a lot of technology around understanding the relationship. Is this something that we focus on here at Precision Lender? Understanding the relationship and what the potential of this deal is beyond the transaction that you're in and having that at the point in time of negotiation.
So all of this only applies to ... If you're the commercial banker, what applies to you are the things that are going to potentially impact your behavior. And that means it has to be delivered at the time when your behavior is going to most matter at the conversations and then the prep for the conversations with your client.
So you're going to see a lot of improvements around that. Here are the product suites that should go with this type of client. Here's the typical balances that go with the client in this industry seeking this type of loan in this time frame. Here are the concerns that you're going to have. Here's even all the way down to the likely competitors.
Jim Young: So it sounds like ... I know because we've had some of these discussions before that AI and banking at this point, we moved definitely from theoretical to practical with this. This isn't thought experiments and the future of mankind. This is how you can use this to do this basically.
George Neal: Correct. You're starting to see a shift in focus from the theoretically possible to the measurable and valuable, which I think is the right direction for any new technology. One of the conversation points that was quite prevalent was what are the appropriate KPIs, key performance indicators for an AI. How do you prove out the value in AI? Even down to when do you retire an AI? You sit there and you look at it and say, "This AI is no longer adding the value that it takes to sustain it. It's fired."
Jim Young: Wow. That is a transition that we're already looking at the possibility of, when is your AI obsolete.
George Neal: We have seen bots retired. Some of them very publicly and disgracefully.
Jim Young: That's true.
George Neal: So this is a problem that we need to address and banks are addressing it. If you've got a bot manager, one of the bot manager's responsibilities is to understand when that bot is no longer contributing to the bank.
Jim Young: Yeah, absolutely. All right, well this has been a great conversation. ROI of AI is a conversation that we'll be returning to on this podcast in future content. But that will do it for this week's show. George, thanks for coming on.
George Neal: It's always a pleasure, Jim.
Jim Young: And thanks for listening. If you want to listen to more podcasts or check out more of our content, you can visit our resource page at precisionlender.com or you can head over to the home page to learn more about the company behind the content. Finally, if you like what you've been hearing, make sure to subscribe to the feed in iTunes, SoundCloud, Google Play, or Stitcher. We'd love to get ratings and feedback on any of those platforms. Until next time, this has been Jim Young for George Neal. You've been listening to The Purposeful Banker.
About the Author
Follow on Linkedin More Content by Jim Young
