York Risk Services Discusses the Complexity of Data and Potential Biases in Algorithms

Kris Merz - Central Ohio Tech Power Player Honoree

Kris Merz

Chief Data Officer

York Risk Services



Phil Haase

Principal – Great Lakes CRM Lead



To listen to the podcast, click here!


Chief Data Officer for York Risk Services, Kris Merz, discusses the complexity of data and potential biases in algorithms.


PH: So, we're here today with Kris Merz, who is with York Risk Services and is the Chief Data Officer and a company that does a lot of processing of claims and works with everything from adjusters to the managed care organizations and really is, it's kind of that, that party in the middle sometimes between the client and the actual services with it. My name is Phil Haase. I'm with RSM. I'll be your guest moderator today. So let's go ahead and get started. So, so Kris, we were talking earlier – you mentioned that you're relatively new to the Chief Data Officer, but you've been a Chief Information Officer. So, when you see those two roles and your role today, you know, how would you characterize the difference between those two from your point of view?


KM: Um, I would say that the Chief Information Officer has a much broader role, right? So, they're responsible for more than just data. So, data is a piece of it, in that it underlines all systems, etc. Whereas in the data space, it allows me the ability and those in those roles to actually really focus on data, and how can it improve processes, procedures – how can it help our end client in the future? So, it's much more of a laser-focused rather than the broader role that you have as CIO.


PH: Great. And I'm assuming you, one of the changes that CIO positions have gone through the last couple of years, or more than a couple of years I guess, is really they've come out of the back room and they've really gotten a seat at the table. Do you feel that that's also the same thing with, from a chief data officer that, as you mentioned, your data is helping people with their processes. Do you feel that that's been a focus that the businesses has accepted as well?


KM: Yeah, I would say absolutely. The role is becoming more and more common. So, you know, five or six years ago you didn't hear a lot about Chief Data Officers, right? Today, it's becoming more of a common role, and I think it's because it's really needed. Um, if you look at the industry of IT, data really is the underpinning of it all. Um, it's taken a long way, a long time for people to recognize that and I know that at York Risk Services, you know, I'm the first Chief Data Officer there. There was a huge recognition of the power that data has here. So there's full support from the executive committee in this role.


PH: Great. And that's great to hear. You know, there's, there's other roles that I think are starting to evolve. You've heard about Chief Information Officers, Chief Security Officers, right? Which is different than what it used to be, which was the person that guarded the passwords, right? And we now have a Chief Security Officer. So, those things have been growing. It's great to hear. And as you talked earlier, you said York has gone through 52 acquisitions over the last few years. I would imagine that would put a laser focus on the challenges of data. And so, as a Chief Data Officer, how have you been working through the cleansing, mastering of those data from not just multiple sources, but multiple organizations with multiple sources?


KM: Yeah. So, so the first thing with anything, and not just this project in particular, but anything you try to do within a business, you have to engage the business. You have to engage those people from each area of the business. You know, we're having discussions around what is our customer, who is our customer? And of course, it's different right now. The view is different by line of business, right? So, we have to have those discussions, um, and come to a conclusion of what is it? And in some cases we may end up, and have ended up, with multiple views of what that is. But at least there's an understanding of how it applies across the organization.


PH: The, the definition of the data beyond an ERD, right? Beyond what the, what the name of the column is, how many characters is in its format, but what it's used for and its context.


KM: Yes.


PH: So, with, within that, the, as, what do you see as some of the trends that are coming into, specifically the data area of IT?


KM: Yeah, I think you, you're probably very familiar with them, right? Machine learning, predictive analytics, artificial intelligence, um, you know, there's, there's a ton of hype right now out there around data and the capabilities, you know, the things that you can do with it. Um, I think that we're, we're up in that high peak of the hype, right? But what's reality today? And so, when you dig behind the covers, right? Sometimes you find, “Well, that's really predictive. It's not really artificial intelligence,” right?

Um, but I think that it's going to continue to grow, and we’ll get better and better at it and, um, in the end, it'll make all the businesses better and make our customers better or to make our, in my industry it, uh,  it’ll make our clients get back to work quicker. You do. I mean, make them more productive, make them healthier. Um, same thing with when you look at claims and the property and casualty space, you know? How can we make that experience for the end client that actually holds that insurance policy better and quicker and easier for them? It's a stressful time. So, I think artificial intelligence and those things will come in and help us do that. They'll do things quicker and better in some cases then we can do, you know, without it.


PH: Yeah. That's great. And you know, one of the things that, just to kind of dive a little deeper in on that is, when we've talked about artificial intelligence in the industry, we've, we've talked about the actual algorithms that we use are the secret sauce behind the keys, right? And in, in with that comes the topic of discrimination and we've had some kind of famous test cases if you will, litmus tests. One of the, one of the most standard ones is the court systems that have used artificial intelligence for sentencing. But, by having a higher than average African American male being arrested, then that affects the statistics, and that affects the sentencing and so on and so forth. And so, there's kind of discrimination in that data and system. And so, as we see that coming forward, how, how do you see that playing out, or is that a topic that you guys are dealing with as well? In the commercial side?


KM: Yeah, it is a topic, and it's a recent topic, right? I mean, I think that as, you know, artificial intelligence becomes more and more popular, people are questioning the results from that, and they should. In some instances, things like race need to come into play, only because it allows us to provide better care. Right? Everybody reacts by race, um, differently, sometimes, to two different medicines and etc. We're learning that. Right? So, in some instances you want that to come into to be a factor, but another way is, you've got to balance that with ensuring it's not a negative factor. So, I think it's going to be a real challenge for the industry and I think you're seeing that being talked about a lot more because it's a recognized problem that we need to address.


PH: And within that are, one of the challenges I'm imagining would be in that is explaining the technical algorithm to that business user about what the effects are of race or religion and something. For example, for looking for someone in foster home, we’re looking for a child that's, and trying to match them up with a foster home - it'd be great to have them with the same race, same religion, right? So they can land softly. But that type of data is very different, and I can imagine that some of the intricacies are…so how are you, how are you approaching that with the business user?


KM: Yeah, I think that we engage in conversations all the time, and we're just in the beginning of some of this journey, you know, since we've grown through acquisitions. But those are topics and discussions that we will have ongoing, and we'll have to get back behind the curtain on what those algorithms are, and we'll also have to do testing on that, right? So, we'll have to tweak those algorithms, see the results to, to try to make sure that we don't have those types of biases in what we're doing.


PH: Great. Great. I think that's great to hear, Kris. And I, and I think that that's really important that the, that everybody's getting a chance to hear in that that is a concern of yours. And so, of course, there’s that  sci-fi movie you read where AI went wrong, right? Where could we have, have altered the bias in the formula, you know? I think that that's, it's definitely a topic I think is important for other, other clients, customers and peers to, uh, to be fostering as well. So, within the technology that you've seen growing, where do you think the most disruptive place is in our, in our future with data?


KM: Yeah, that's a, that's a really tough question. I’m not sure I have a really good answer for you, because there are so many possibilities with artificial intelligence. Um, you know, you brought up the science fiction. Um, you know, you do have people out there that are very fearful of that, right? We found, we just talked about biases, right? We have found that artificial intelligence, uh, while it’s learning right, and, and adjusting, that those can be influenced.

So, I think it's a hard question to answer right now. I wish I had a crystal ball for the future. Um, I don't, but I do see it playing in nearly every industry, you know, for a long time. Um, insurance, or the space that we play in, was not affected. Well, you see the rise of what's called InsureTechs – lots of investments over the last couple of years. And that, a lot of those are very positive and we've got to be careful about where we take those.


PH: So Kris, we've, we've talked a little bit. originally, you know, in the beginning a little bit about the difference between the CIO and Chief Data Officer. Then we talked a little bit about where your focus is and some of the issues you have, you're focused on in your company around data. So, one of the questions I want to circle back to is, within the role of chief data officer, where do you see that role evolving to within your industry or, or just the industry in general?


KM: Yeah, I would say I think that in many companies the Chief Data Officer, um, will become more and more important to the success of the business, you know? More striving at York to really become what we call a data-driven company. Um, so let's look at our business with data in mind.  You know, as we do things, what is the underlying data and information that we're collecting, and how does that help us, and how does it help our clients? Um, so I think that the role is going to expand, and it's not just in insurance. You're seeing it across the board. Um, in many companies.


PH: Yeah, so, so wherever a company is looking to be more data driven, then it'd be hard to do that without someone on the board whose job is And, t hat's an interesting point of view. I hadn't really thought all the way through that that as you become more, as your company pivots more around that item, you're going to have to build an organization under that, that pivot. So, within your organization that you're building, what are some of the key positions that you have under, under your purview that are, that are helping you in your mission?


KM: Yeah, so, so, we're building the organization out. Since I'm new there, there of course wasn't an existing, you know, organization. So, um, you know, you need those people that are tactical, that do the day-to-day. Um, you know, things like data architects, but you also need to have those roles of data scientist, right? People that can take and read that data and leverage it.

You also need people that are able to kind of bridge that gap in between, whether you want call them, you know, business analysts – I think they're really more of a bigger role in data than that, that understand the business and how does that tie with the data scientist? You know what I mean? And what, identifying what problems are we trying to solve? And in some cases, having those people be able to dig into the data and, without being told of a problem, find the correlations and say, “Hey, you know, I'm kind of seeing this pattern here. How does that help us? How does that help our client”?


PH: So, for that data scientist role, which is relatively new-ish, right? It might have been called something else in the past, but now it's really, you're right, kind of been identified. Do you feel that you're growing those, or do you feel that you're hiring those out of college, or are you hiring them from a different industry?


KM: We're really going to have a mix, uh, what I see as a mix. So, I would love to get some people that are just right out of college that have, you know, um, bachelor's degrees or master's degrees in this space. Um, I think that, um, unique view of not been tainted with prior information is really good. They don't come in with any preconceived notions. They tend to see things. At the same time, we do need those experienced people, right? Um, and people that know the industry, so I really see a mix of people within that space.


PH: And so the, I'm assuming that you're putting training together, uhm, experiential training for those people that are already in your industry and experience to get more into the data, right? You're educating them on the data. And of course, the people coming out of college – you're right, they're not tainted with those previews. They’ve got other items that they're, that they're working on, right? There's other things, you know, like, “Hey, this is a job now. Um, it's not just a class you go to Wednesday afternoon, that this really is, you know, that nine to five.” So, um, that's great to hear. I, I know that we've looked at a bunch of items around, you know, where in the industry, what are the needs within the information technology industry, and what colleges are universities are providing those in what ways? Um, and are there other institutions as well as universities, that are providing those talent bases for us? Um, so within your efforts to bring together these 52 companies and your other data pieces themselves, is there specific technologies that you're using or focusing on around, like, the mastering of the data?


KM: Yeah. So, you know, there's many, many companies out there that have different tools that you can utilize to help with data quality and data integration and data mastering, and you'll hear the term “data stewardship.” Do you know what I mean? So, there's lots of different tools out there. Um, you know, we have just recently gone through an evaluation with those and selected some of those tools that I think will really help us because they have built into it, um, some of the abilities that we need, without us having to create it. So, some of the matching algorithms, you know, so, some of that AI, so to speak. You know, they may have clients or customers that in that are in our systems 24-50 times. You know, they may have different names, um, that are similar but different, different locations. So, using some of those technologies, we're able to get to a quicker view of a single customer by using those. So, definitely looking at some of those tools. Like I said, we just recently selected one. Um, we're also looking to build out, um, the first data warehouse. Here at York, they have tons of data, they have tons of data marked that has been, been pretty effective over the years, but very isolated, you know, based on each individual company and, and acquisition. So we'll be pulling all that together.


PH: And do you also have or lead within, within York, like, a data governance team or a data governance position?


KM: So, it's one of the things we're building out. Absolutely. So, we're not there yet – it's a, it's a team that we're putting together as we build out all these different functionalities and single versions of the truth of customer or product or et cetera, m, there will be a governance component that we put in place as well.


PH: I know that that's usually the, kind of the second step, almost, that happens, you know. So, when we've looked at the Chief Innovation Officer, then they put together an innovation council right? The Chief Security Officer, and then they put together a security council and I know, I'm imagining with the Chief Data Officer would then have a data governance council, right? That we can start to adhere to the rules of, of data throughout the organization. I imagine that that's kind of one of those second steps.


KM: Yes it is. Um, we're putting it in, we've got a lead, right, on building, starting to build out the single view of customer or et cetera. We are just now bringing up that governance board. Um, I think it will be critical to us because, you know, security of data is critical. The privacy of that data is critical. We have to understand where that data's being used, by whom, and what for. Right? So, that board will help with all of that. That governance body will help us with that.


PH: That's great. I think that really helps. Um, you know, kind of what I'm saying, kind of round up my part of my curiosity around how the chief data officer is different too than the CIO in, within your organization, what are the pieces that you're kind of checking off the list in any of those orders for that.

So, the…to close up, one of the questions I have for you is, within, this Columbus market is really becoming a hot IT area for technology and so, you know, where do you see yourself, especially within your, your view of the industry, where do you, how do you see us getting to that next level here in Columbus around data or technology in the community? Do you, where do you see that going?


KM: Yeah, I see more and more of the community pulling together, quite honestly. You have things like the Columbus Collaboratory, right? That has a whole analytics program that they started out with and have continued to expand and are now sharing some of those findings and information. You have things like the CIO Forum that's in place. Um, there's definitely much more communities around data than there ever has been. I think that will continue to grow. Um, each of us are trying to tackle the same problems. Um, and I think that the more sharing that you do of how you're tackling those things, especially critical issues, I mentioned privacy – it's just, it's absolutely critical today, right?


PH: Right.


KM: Um, so as we have those conversations, I think it helps each of us, and so, I think it will continue to grow, especially in the data space and have much more focus.


PH: That's great. Yeah, I think it's a really exciting time to be in IT and in Columbus, Ohio. Well, great. And thank you so much for your time. This is Phil Haase and Kris Merz, and to learn more about us, why don't you go ahead and visit comspark.tech, and goodbye, until next time.


To learn more about sponsorship opportunities for 2019, contact Michelle Ziegler at michelle.ziegler@venuemag.net