December 08, 2020

Article at IBM Systems Magazine

View original

iTalk With Kent Milligan

In this episode of iTalk With Tuohy, host Paul Tuohy talks with Kent Milligan from IBM Systems Lab Services.

Paul Tuohy: Hi everybody and welcome to another iTalk with Tuohy. And I know I often say it's a delight to have a person with me on these, but today I can truly say this truly, truly is a delight. So hello, Kent Milligan.
Kent Milligan: Hi Paul.
Paul: So some of you will remember Kent from a few years ago. Kent worked for a long, long time with [IBM] Lab Services, speaking at many, many conferences, writing many, many articles out there. And then Kent left and went away to the wonderful world of Watson—and, so which is one of the things I want to talk to Kent about, but the great news that I want to share is that Kent is now working—back working on IBM i again, is back in Lab Services. So, great excitement. So Kent, I think the thing I'm just dying to know, please, about working on Watson, what—what that was like and what it is you were doing, and your experience on that. So, in your own words.
Kent: Okay. I mean I think the highlight for me kind of overall change into Watson is the project that I worked on the most was a product known as Watson for Clinical Trial Matching that basically speeded up the process that it would take for an oncologist to, when they have a patient who normal cancer treatments aren't working, make it really quick and efficient for them to find out what clinical trials are available that would be a good match for their patient. Because as you might imagine, oncologists are busy seeing cancer patients all day long, talking to them about their treatments, reviewing their treatments. It was really kind of almost in their spare time was the only time they had available to keep up on the latest clinical trials that were available. Some of the IBMers that were first involved in the solution, they met with an oncologist and he basically had a white board where there were like ten clinical trial he would keep track of as possible options, when in reality there's hundreds of clinical trials that might—you know might provide a solution for their patients, but they just didn't have time to read all that literature on their own. So you felt really good that you know the technology you're working on was you know giving cancer patients a chance to find a cure for their condition.
Paul: Yeah. It's—I think it's that area of—it must have a been great feeling sort of knowing that everyday that when you would go to work of the like the benefit of what you're doing.
Kent: Yeah, there's just a little bit more personal feeling of pride where you know when I was working in my prior life as an IBM i person right, I might help like an insurance company or a trucking company, help their applications run faster, right? I mean there's some level of satisfaction in that, but nowhere near the level of knowing that you're possibly helping improve the—somebody's health and their everyday life by what the technology is doing.
Paul: Yeah. So when you're working—when you were working on something like Watson and that Kent, I mean is it a very different working environment from-from what we're used to in the usual sort of you know software development maintenance type environment and that? Is it completely different or is it basically the same, just different type of components that you're working with?
Kent: I'd probably say a little bit of both. You know the development processes and the approach development were quite similar, but the thing that you had to kind of adjust your mind to and we had to adjust customer's mind to is when you're working with artificial intelligence solutions and machine learning type solutions, it's not like an OLTP application where it's going to be right you know 99.9% of the time, right? A human and machines trying to simulate human brain or whatever, you know it's going to be right most of the time.
Paul: Yeah.
Kent: So that's kind of a mindset or paradigm, you really had to kind of move past it. Well I can't expect this to return the answer 99.9% of the time correctly—so it took kind of some adjustment to do that and then trying to come up with insights of when do you press the development team for a fix vs. say well this is kind of a fringe thing and it's not super clear, so I wouldn't expect a piece of software like Watson to get that right. It's not going to provide value enough to—to bother the development team to adjust some of the algorithms to get that right.
Paul: Yeah so—so having worked sort of in the bowels of it, are—are you—has this whole area now of AI just become passé to you, or are you still mightily impressed by it?
Kent: I mean it was definitely impressive, what it could do. I think the other I gained a new appreciation for is that you need people with, you know, industry expertise or subject matter experts to really help it move along to the next level. I mean you could have great computer scientists behind the scenes cranking out algorithms, but they need to be working in tandem with the subject matter experts to really tune it in a way that's going to provide value to the people in that industry like the healthcare industry.
Paul: Yeah. It's one of the things that I found fascinating that the—the very few people I know who've worked in the area, I think everybody has said the same. And anything I've read on that is that really what drives it is as you say those subject matter experts, that they're the—it's their knowledge of whatever the intelligence is on is really what is key to that development, as opposed to programmers or developers or as you say people writing algorithms.
Kent: Right, yeah. It's really easy to write software that—I mean I spent a lot of time focused on the software that would read and try to understand doctor's notes for patients on their visits and stuff—
Paul: Oh wow.
Kent: It's really easy to write software that reads that and can parse it and categorize it, but to do anything with it, you need the subject matter expertise. I think it was about two or three years ago we were working on training Watson to understand bladder cancer and I was on the phone for like a half hour with two medical people who were explaining to me the different parts and components of the bladder, and it was just like "what have I gotten myself into here [laughs]?" I think I'd much rather understand a Visual Explain optimizer plan than understand how the human bladder works, you know type thing. So it was kind of a surreal moment or whatever that I was in the middle of this conversation like "yeah I wouldn't have guessed having this conversation ten years ago" or whatever.
Paul: Yeah. So—so do you—for the future of AI, do you think is it just now at its really, really early stages and it's going to become enormous, or is it something that you think is yeah, it's got a little bit to go but there isn't going to be much more to it?
Kent: I think it's maybe further than the early stages. I think—I think it is definitely taking steps and the industry is learning on how to best utilize that type of technology, you know with real customer solutions. So I think there are definitely improvements that can be made and that I'm sure the teams working on that will make.
Paul: Yeah. So I mean just be thankful by the way when they're talking about artificial intelligence that it isn't based on something like mine. If you want a hit rate of like, you know successful or being right about one out of ten times—
Kent: Right.
Paul: The bar would be—the bar would be very low.
Kent: Right. Yeah.
Paul: So listen to more important things though, Kent: so as I said, sharing at the start that great news about you being back with Lab Services, back working on IBM i, back working on the database again. So I suppose the first question really is—well sorry: I'll ask you two questions. You may want to answer them together. So—so one is going to be so what exactly are you going to be doing, but more importantly, why? Why have you gone from working on an area like AI and Watson and all, and come—and come back to the wonderful world of IBM i?
Kent: Yeah so when I first left for Watson, it was actually in the Watson lab services area and it was kind of more targeted around customizing Watson-based solutions for customers. So you had a lot of interactions with the—the end users and the clients to figure out how to best customize, but in the last 2+ years working in Watson I moved purely into a back end development role, kind of working more on the—the core product that before I was customizing more at the end user level. Now I was more working on that core product development team. It was nice to return to some of my development roots. I mean I spent the first ten years of my career with IBM, you know working on the development team here in the Rochester lab. So I did kind of enjoy being back part of a real development team and the development process, but after two years I kind of also figured out that you know I was missing that customer interaction and the satisfaction that comes—whether it's kind of a one on one interaction with a client or you're presenting a set of customers like at you know the RPG & Db2 Summit—of knowing that you've explained or helped them use the technology that's going to make their job easier going forward. And I—you know I'd had some conversations you know with my Watson manager of how we might accomplish that, but then you know an opening came available on the Db2 team and I started to really think that yeah, I really do miss being able to help customers directly solve problems as opposed to the development role.
Paul: Yeah and okay, come on be honest, Kent: It's all because we're such nice people in IBM i. That's truly what you missed. Come on [laughs].
Kent: Well I mean—I'll have to agree with you there. The IBM i kind of community is pretty unique. I mean I'm sure there's a lot of great people on the Watson side but you know that product and community just didn't have the length of time, right—
Paul: Yeah.
Kent: Or as many clients involved in it that with those relationships. So I have to admit it has been fun to you know reconnect both on the IBM side of the things and then the external side of things with all the IBM i people, because I had stayed kind of connected to some people but not to this degree. So it has been really nice to kind of come home and more directly connect with people.
Paul: Okay so have you reconnected with any of your old favorite customers that you—that you used to deal with?
Kent: I don't know. I'd have to say I was gone five years and I'm not sure I remember who my favorite customers were [laughs].
Paul: Okay, you've just—three people have just started crying, Kent going he doesn't remember us.
Kent: Well I told you before they were filling my mind with all this cancer and bladder information, so there's only—only so much room in my brain to keep track of everything so.
Paul: I hate to say it but the next time I look at Visual Explain all I'm going to see is a bladder. You've ruined it for me [laughs].
Kent: Yeah, yeah.
Paul: So—so Kent just to I think—sorry, there is still so much that I want to ask you, but the—I think we will wrap this up. There's—I know we were chatting briefly just before and before we started to record, and I know from past experience that I think—I was trying to think actually was there ever one conference where I met you where at some stage you didn't disappear for an evening because you were going to basketball game, a football game, a soccer game or some sort of a sporting event. I know you're a massive sports fan, so have you been finding it difficult the last few months with no sports being on or not being able to get to them?
Kent: Yeah. The first—the first couple of months were a big—big change for me. As you said, you know that's kind of my hobby either watching sports or attending them. Then you know we got them on TV but then I was really worried about this fall when American football looked like it might not happen because our daughter, who was playing college soccer has now graduated from university, so this is going to be our first fall without soccer and if I didn't have online soccer and online football to watch, it was going to be kind of a pretty depressing time. So I was glad that that came through.
Paul: Yeah, I mean this could be terrible Kent. You might have to, I don't know, talk to the family or something [laughs]. It would be terrible.
Kent: Yup, yeah.
Paul: But it's—but so I'm sorry. Has basketball restarted in the U.S.? Did they-?
Kent: Yeah they played. They ended up playing the professional season this summer and it just ended a couple of weeks ago. As far as I know the college basketball is going to start—but I mean the good part is that I have that to balance me out and all my SQL performance and best practices work that I'll be doing with Lab Services, I won't get too carried away from that because I'll have something else I enjoy at night to kind of say hey, I can put that aside for now so.
Paul: Yup which is—which is the whole point of it.
Kent: Right, right.
Paul: So Kent, I really look forward to when we meet in person again, and thank you for taking the time to talk to us.
Kent: Yeah, I look forward to that as well. It was good catching up with you, Paul.
Paul: Okay. So that's it for this iTalk, everyone. Tune in again soon for the next one. Bye for now.

Subscribe to IBM Systems magazine, Power-Systems