The Ohio State University Discusses the Potential Dangers of Biased Algorithms and Keeping the End User's Context in Mind
Diane Dagefoerde - Columbus Tech Power Player Honoree
Director of Market Strategy - Central Ohio
To listen to the podcast, click here!
Deputy CIO for The Ohio State University, Diane Dagefoerde, discusses the potential dangers of biased algorithms and keeping the end user's context in mind.
Hello, and welcome to the comSpark podcast, where you will get to meet today's technology thought leaders. To learn more, visit comspark.tech.
SG: We're here today with Diane Dagefoerde, who is the Deputy CIO at the Ohio State University, an organization that spans everything in education with over 29,000 employees and over 64,000 students among the five campuses. My name is Steve Gruetter. I'm with Expedient. I'll be your guest moderator today. Diane, let's get started.
DD: Sounds great, Steve.
SG: You have been with The Ohio State University for over 20 years and in your role as Deputy CIO the last four years. This is a situation where most CIOs only last 18 to 24 months on the job. To what do you attribute your longevity and your position with the university?
DD: Steve, I love what I do at Ohio State. I connect with the mission, you know, our mission as a research institution, also teaching and outreach, everything that we do at the university every day. It makes me really excited to be a part of it. I mean, for researchers, the technology systems, the core IT solutions and services that my team runs. We have an opportunity every day to help researchers create new knowledge, new knowledge that can solve a problem in the world. Our technology supports students as they go through their education, graduate and go onto their careers and the rest of their lives. Ohio State's a great corporate citizen, both in Central Ohio and nationally, and it's really fun to just be a part of that. Every single day is a new challenge. I learn something new every day. I work with a great team of people – I can't imagine working anywhere else.
SG: I tell you, I read my OSU Alumni magazine every month, and I am astonished at the level of, of the capabilities and what the students are doing now. It's really a tremendous level of activity as compared to how it was 30, 40 years ago, and it's been a tremendous uptick in what the university is accomplishing.
DD: It's amazing. The student employees that we’re fortunate to work with in my office, um, oftentimes we recruit students who are in the undergraduate data analytics major. They come and they work in our business intelligence analytics and reporting team, and we give them messy problems. Let's see what they can do with this, and they come up with amazing solutions. And these are students that work with us during the academic year and then in the summer they're off doing internships with Google or AWS or Facebook or whatever. I mean, they're really talented. It's amazing the talent that we have at Ohio State today.
SG: I am able to come back to my alma mater twice a year to help teach Bruce Barnes’ eight MIS class. And uh, the questions that come out of the students are significantly better, often, than the questions that come out of the IT professionals that we work with every day. The students are much less jaded.
DD: Yes! They’re lots of fun.
SG: So, a part of your role as Deputy CIO is, uh, working with emerging and disruptive technology, and of course the university is world famous for that, in your opinion, what is the most exciting disruptive technology that you're working on that will impact our lives?
DD: I think artificial intelligence, especially as we apply it to services and the way we use it to automate decision making. That's the technology that, to me is, the most exciting, and also the scariest. So, it's exciting because, you know, technology, if we do it right, it amplifies and extends what we as human beings can do. You know, technology, there's a lot of talk about automated decision making and are we replacing people or are we, are people losing jobs? I don't see it that way. I see technology and automated decision making – like call routing for example – it's something that enables people to then elevate what they do and really leverage emotional intelligence, curiosity, you know, things that humans are really great at.
So, I think it's exciting, but like I said, it's also a little bit concerning. An algorithm, algorithms behind these automated decision-making systems, algorithms are only as good as the trainers that they had. They're just opinions written in code or expressed in code. I'll give you an example, so, working with students, we have to be concerned about, you know, the whole student, including their health and sometimes their mental health, and one of the things that we're looking at doing is collaborating with a group like the National Suicide Prevention Lifeline to make that available through the Ohio state app.
It's something that students, most students at Ohio State, over 95 percent of the students at Ohio State use the Ohio State app every day – sort of a lifeline for them for everything. So, why not mental health? But, one of the things we've come to understand is that, in suicide prevention, it's important for the caller to be paired with someone who's from the region that they're, that the caller is also from. So, you wouldn't want to pair someone who grew up in the south perhaps with someone who grew up in, with a call support person who's in Alaska, for example. There's just not enough shared context there.
SG: Makes sense. Sure.
DD: But the way the algorithm that the hotline uses today is based on the caller’s area code. Well, we have students from all over the country, so, we'd want to make sure that they are connected with a local resource here in Central Ohio, not a resource in their hometown, so that kind of decision or assumption that's built into the code, “Oh, well if they're calling, you know, look at the caller’s area code and let's use that to figure out which suicide counselor to, to connect them with,” that's a faulty assumption, because it assumes that the area code relates to actually where the caller is. Where a location based service might be a better assumption.
SG: It makes plenty of sense. Makes plenty of sense. That is a great example. Do you have any other examples in there where an assumption will corrupt a code, if you will?
DD: Sure, so think about nontraditional students at Ohio State. So imagine you're a single mom, and you have a couple of kids. You're working two jobs, maybe you've gone through a nasty divorce and that caused you to take on a bunch of debt, maybe affected your credit in such a way that you no longer have access to a major credit card. You know, it's not, it’s something that a lot of people have to deal with on a daily basis. So we've got an app. If we were going to make it possible for someone to pay their tuition bill, let's say, using a credit card through the app, that's great, but that doesn't help this person. And the reason it doesn't help this person is, they don't have a credit card, and maybe they even need payments, and they're working two jobs, so that if we don't have it in the app, now they have to go and contact the office, and they have to contact the office during open office hours, but this is a person who's working two jobs, so it's really hard for them to do that.
They've got to take time off work and lose money out of their paycheck just to go figure out how to pay their tuition bill, because we don't have an option for them in the app. Why don't we have an option for them in the app? Because we didn't think about it. Because we didn't think about this person's context or this person's situation. That's the kind of thing I'm most concerned about in this world of artificial intelligence and automated decision making. We as humans really need to bring our curiosity around, about all people to bear when we design these algorithms. And if we don't think about a use case or context, it's likely we'll miss it in the code, which could end up actually hurting the population we intend to help, as in the case with the single mom.
SG: I think it's great that you are addressing these issues and trying to make solves for those exceptions out there and really considering every solution, no doubt. You had mentioned previously about a, some books that you thought, certainly, that you have a true belief in.
DD: Right. There's one book that I've asked my entire team to read. It's by Kathy Sierra and it's called Badass: Making Users Awesome. And what I like about this book, it's a, sort of a graphic novel written by a developer herself. Um, but what I like about it is, it helps in, in her novel, in her book, Kathy helps us understand this idea of technology, its purpose being to make people awesome at what they do. You know, we're not here to make students be better typists. That's an old example, I guess. We're here to make students be better data scientists. And so, technology's got to help them do that, not get in their way. And you know, when we as corporate IT, roll out upgrades to software that everybody across the university has to use because they have no choice, we really need to take into account what that does to our researchers, our students, our staff and their ability to get their job done.
So, if you think of a faculty member who is maybe further along in their career, has just now switched from an analog phone to Skype for business phone. Now they're in class and they're in the middle of their lecture and their laptop rings, and they have no idea how to turn it off. You know, this, for them, is not progress. So, we have to keep in mind always the user and their context and how we can enable them to be more awesome with technology solutions we put in place, not have technology get in the way.
SG: Make it easy.
DD: Make it easy.
SG: That you've been a part of the Ohio State University for 20 years – obviously you've kept in tune with what's going on here in Central Ohio. What is the best part for you about working in the technology community in Central Ohio?
DD: By far it's the people, the community in Central Ohio. The tech community is collegial. It's dynamic. I know that I can, if I have a problem that I'm trying to solve at the university, I can pick up the phone and call any tech leader in Columbus and number one, they'll take my call, and number two, they'll help me. That's rare. I don't think you see that in many communities nationally, but we've got that here in Columbus.
SG: I don’t think anybody collaborates like we do here.
DD: I don't think so either. It's really great. It's something that makes Central Ohio, working here, such a joy.
SG: It is also the very positive aspect of OSU, where it touches so many people's lives in Central Ohio, and in such a positive manner, that when the call comes in from OSU, generally the first thing people want to do is, they want to help!
DD: And then they ask for football tickets.
SG: Yes, they do. Yes, they do. So, if there's something that we could be doing better in Central Ohio in the technology community, what do you think that would be?
DD: I also think it's people. We have a talent problem here in Central Ohio, meaning there's an insatiable demand for tech talent, with all, from all kinds of areas – programmers, data scientists, but also technicians that provide support to overall uses of technology. We've got to find a way to attract people to Central Ohio, make sure that the students that we educate here, not just at Ohio State, but other universities in Ohio, stay here in Ohio. We need to make it possible for a person to be able to have a career in technology in Ohio, but not necessarily have to live in a city. You know, better connectivity in the rural counties in Ohio to give people choices to have the lifestyle that they want, but also be able to have a great technology career.
So I think we need to do things to make it, Ohio a more attractive place for talent, and I also think we need to do a better job of growing our own talent and maybe doing that a little bit differently. Of course we have higher education, but there may be some other approaches that we might need to consider to grow our talent in other ways.
SG: I have not heard an organization in Central Ohio that has said that we have enough talent. I know that our organization is hiring. I know that dozens of organizations are hiring, and it's a great place for us to be right now – essentially zero percent unemployment in our space, but it's a, it's still a problem. It's a good problem to have, but it's still a problem. So, thank you Diane. So much for your time. This is Diane Dagefoerde, of the Ohio State University, Steve Gruetter of Expedient, and to learn more about the podcast that we're doing, please visit comspark.tech. Goodbye, until next time.
DD: Thanks, Steve.
To learn more about sponsorship opportunities for 2019, contact Michelle Ziegler at firstname.lastname@example.org