Innovating at the same time everyone else is innovating is hard, but U.S. Bank seems to be finding its way around.
Headed by its fearless leader, Dominic Venturo, the bank’s Innovation Group has been at the epicenter of innovation in financial services across various tech sectors. The group has a lot going on at all times, Venturo notes, but AI initiatives seem to be of particular focus at the moment.
Last month, the bank posted a job announcement for an innovation leader on its website, signaling an increased emphasis on the technology and its applications.
Bank Innovation caught up with Venturo to talk AI and other projects within U.S. Bank’s Innovation Group.
(Responses were edited for length and clarity.)
Bank Innovation: About that job posting. Are you creating a separate AI-focused team within your group?
Dominic Venturo: We have actually been working on AI and machine learning within the Innovation team for a few years now. We have done a number of interesting experiments and pilots, and, as you know, there is a broad scope of topics that can fall into that. So, with what we are doing now, we thought that there is enough going on that we want to have a dedicated group within the innovation shop to explore applications in various businesses…
This is not unlike the way we have gotten into mobile payments space — a test and learn kind of an environment. So we want to bring in a leader to help coordinate things, also provide strategic leadership to the direction that we go, much with the existing team, but with a growth expectation over time.
BI: What are some of the most interesting AI projects you have completed so far?
DV: We started pilots with customer service interaction some years ago. We began with voice processing, so providing customers an ability to have a verbal conversation with an intelligent agent, and get answers. What we found back then was a lot of structure was required around how inquiries were made. So part of the problem there was that the system was not “smart enough” to understand the full context of English that’s being spoken, and that’s okay, but it was a limitation at the time. Now, the technology has advanced quite a bit since then.
Recently, we completed a pilot of a text-based computer chatbot that you would interact with through a chat window, or texting, or [an] IM window. The pilot has just wrapped up, and we haven’t announced what we plan to do with it next. So far, our thinking is that the ability to allow an individual employee to have fast access to the customer information at their fingertips is very valuable. Employees can get real-time information about any one of our complex features, or offerings, allowing them to have almost an encyclopedic knowledge.
There is still a lot of training required of this assistant to deliver the same kind of service that a human would deliver, even a lightly trained individual, like a new employee.
BI: What about AI and machine learning for underwriting decisions?
DV: I think that’s an interesting one. One of the challenges we have been wrestling with [in] cognitive computing, deep learning and AI as an industry is that these tools can make a decision based on the algorithms and based on information from consumers. However, those tools may not always be able to explain how they arrived to the end result; sometimes it can’t tell you how it got from Point A to point B. It gets hard, because you have to be able to be confident about the decisions you are making, and those should be predictable and repeatable.
But it can be done; it’s just very early days for this technology. One of the things you have to do is recreate an environment to test and run those processes to fully understand how they work. And the key may actually be in having more data inputs, not a different computing engine.
BI: What are some other areas you are looking at, besides AI?
DV: One of the newer ones, and it’s not as sexy as AI, but coming soon is a real-time B2B payments platform. The underlying infrastructure will play a lot of roles in the way financial institutions do what they do. So it’s an industry level initiative, and we expect a lot of new capabilities to be unlocked.
We are very interested in biometrics and authentication from the security perspective.. One of the challenges with mobile devices has been long complex passwords, while biometrics also have an added layer of security.
And finally, distributed ledger technology. Just last week we’ve invested in [the] R3 consortium, which we initially joined as a member bank. I do think a lot of projects from that sector will translate into real-life applications soon.
BI: How do you attack new technologies within your team?
DV: We literally scan the globe every single day to see what’s new and different.
We work with a lot of partners, many are tech companies. The team has grown a little bit, we are currently about 30 people, and our structure allows us to have pure R&D focused folks, that oversee products from research to testing to deployment. That’s what we did with emerging tech in payments. Contactless payments started in our lab, and they are no longer in R&D. Distributed ledger technology is a good recent example, too.
Part of our thinking is practical: is the technology scalable? Is it viable? Do any lenders use it? And then we try to make an estimate of where it is, and what’s the viable path. We are almost always wrong with those estimates, but we are right directionally. Then of course there is the secret sauce of how we think about a technology and its implication.