Alex and Amy talk with Montana CIO and co-chair of the new NASCIO Generative AI Working Group, Kevin Gilbertson. We talk about how he's discussing generative AI with his team, state agencies and other CIOs, the benefits, challenges and things to keep in mind. He also answers the age old question: Star Wars or Star Trek?
Find our transcript here: https://www.buzzsprout.com/704052/episodes/13365163-how-states-are-thinking-about-generative-ai-with-montana-cio-kevin-gilbertson
Alex and Amy talk with Montana CIO and co-chair of the new NASCIO Generative AI Working Group, Kevin Gilbertson. We talk about how he's discussing generative AI with his team, state agencies and other CIOs, the benefits, challenges and things to keep in mind. He also answers the age old question: Star Wars or Star Trek?
Find our transcript here: https://www.buzzsprout.com/704052/episodes/13365163-how-states-are-thinking-about-generative-ai-with-montana-cio-kevin-gilbertson
Thu, Aug 24, 2023 10:34AM • 18:17
SUMMARY KEYWORDS
AI, state, generative AI, kevin, data, gpt, references, pocs, cautious, robot, provide, tool, star wars, cio, different use cases,
Kevin Gilbertson, Alex Whitaker, Amy Glasscock
Amy Glasscock 00:05
Hi and welcome to NASCIO Voices where we talk all things state IT. I'm Amy Glasscock in Lexington, Kentucky.
Alex Whitaker 00:11
I'm Alex Whitaker in Washington, DC Today we're talking with Montana State CIO Kevin Gilbertson.
Amy Glasscock 00:17
Kevin is co-chair of our new generative AI working group. And today we're talking with him all about this new technology that's on everyone's mind. Kevin, welcome to NASCIO Voices, and thanks so much for joining us.
Kevin Gilbertson 00:28
Well, thanks for having me.
Alex Whitaker 00:30
So Kevin, before we dive into generative AI, tell us briefly about your professional background and how you ended up at state co in Montana.
Kevin Gilbertson 00:38
Yeah, sounds great. This is my first government job ever. I've been in private sector my entire career. And I've got a little bit of an eclectic background, you know, I actually started as a technical writer, became a computer programmer did some time in sales, and a lot of, you know, professional services type of work. I had worked with the governor before at a previous company. And as a matter of fact, that was actually recommended by a good friend, who also worked at the company when they were brainstorming for the role. So it all came together. And it's been great ever since.
Alex Whitaker 01:10
Yeah, awesome. I think I don't think we've ever interviewed a CIO, who has said, Oh, this is my straight linear path. And everybody always has a really interesting background. So thank you. So as we mentioned in our introduction, you are co chair of NASCIO's new generative AI Working Group. Talk to us about your understanding of generative AI, and how would you explain it to someone who's not familiar with the concept?
Kevin Gilbertson 01:31
Yeah, so I'm going to do this with a couple of I, you know, it's anticipated this would be coming. But I saw I plugged it into a generative AI tool to see what they would say that I should say this. So what they told me to say is, and I won't say which one is that represents a major evolution in AI capabilities that uses machine learning to create realistic new content, like images, text, or videos from small prompts, has a tremendous potential business applications and could transform how states operate. And it can help with content creation, and then it does go into some risks and things like that, which is not a bad explanation. But I think, you know, for a lot of folks, this is a little confusing. And so, you know, when I have been talking to folks that don't know it, I you know, assure them we're, we're not entering into Terminator territory here, these things cannot think all they are is prediction machines that have been fed a whole lot of data. So they can predict which pixel which word to put in there. That is similar to the context of the you know, the question or prompt being asked. And so, really, that's all it is at the end, but at the same time, you know, once again, like the Gen AI version, it is great for image types of videos and getting work done.
Alex Whitaker 02:40
Me I think we need to start having AI write, some of our questions for the podcast.
Amy Glasscock 02:45
Right? Yeah. You know, it is funny, I did ask ChatGPT before this interview, what questions should I ask a CIO about generative AI? So one of these questions may have come from ChatGPT actually, I just don't remember which one. Kevin as a sort of follow on to that. My grandmother is 101. How would you describe it to her?
Kevin Gilbertson 03:13
Yeah, I'm not sure. Essentially, a computer tool that creative? You know, I'll have to think about that one. [laughing]
Amy Glasscock 03:22
Okay, [laughing] we'll come back to that one! Okay, so tell us about how you are talking about generative AI to your team? And how are you encouraging them to use it? And then on the flip side, how are you cautioning them of any potential risks?
Kevin Gilbertson 03:39
Yeah, that's a great question. So I am encouraging wherever I can. And wherever it makes sense, obviously, there's a lot to be cautious about. And so within my enterprise architecture group, we have put together a AI document to provide guidance to the different agencies. And so that's still in draft form, which we're putting out. But in general, for example, I asked my comms person if she was using any generative AI for helping to generate the communications that are coming out of the office. And I think that she was worried that I was going to be upset if she did. And I basically said, No, you're crazy if you don't, you know, type of a thing. But you know, I'm cautioning folks to say, you know, only use it for topics that you're already familiar with, or it's not a complete expert on because, you know, one of my very first uses, I asked it, to provide an answer to a question that I had. And then I said, well give me a list of references in APA format for where you got this information that I could look into that would support, you know what you've said, and so it gave me a beautiful list of references. And then I went and looked for all of them, and I couldn't find them. So I assumed it was my fault that this was very early when I started using it. So I went back and said, Are these real references? And ChatGPT responded to me saying, No, I just made them up to fit the context of the answer that I gave you. And then I was able to say, provide me some real existing references that would, you know, do the same thing. And it was able to do that. But, you know, that was probably my second use of the tool ever. And so obviously, it became a caution for me that I pass on to others. So you know, only do it if you're familiar with it or an expert, because it is a great way to get a head start on something, but be cautious that you're only going to be able to use maybe 60% of it, you're going to have to do some rewrites and make sure that it's in line with what you're looking for. And then be prepared to do some deep dive research into what you're getting back if you're not. And so, you know, from that perspective, it will still end up saving you time, but you have to be skeptical every time you use it.
Amy Glasscock 05:45
Yeah, that's a good point. I was talking to my cousin who has a PhD and in basically, computer science, but he specializes in artificial intelligence. And he was telling me that, you know, he was playing around the ChatGPT asking some questions. And it came up with a paper that he's like, Why have I not heard of this academic paper? I know, the people that wrote it, you know, this sounds like something I'd really be interested in. And of course, it was fake. You know, it's a good example. It's kind of good for brainstorming. But you do need to know, you need to be able to recognize the false outcomes.
Kevin Gilbertson 06:19
Yeah, well, once again, it's just predicting which word should come next. And if it found those names, and other publications that match, and this is something similar, you know, that's what it's gonna give you. It's not really out there to, you know, tell you truth or lies. It's just, that's all it's doing.
Amy Glasscock 06:36
Yeah, yeah. So I'm curious how so that's kind of like how you're talking to your team about it? How are you talking with the agencies that your IT department serves about using it.
Kevin Gilbertson 06:48
So at this point, you know, we are asking them to be cautious, and I'm really working with folks that are raising their hands to be early adopters, we do have a few of those that we're looking to do some POCs on, we don't have a lot of use in the state yet, not like some other states that have have gone out and have already started some projects with it. So we're still kind of in the investigation stage, you know, we're looking at different things. Like, if you're planning a trip to Montana, when are you coming in? Where are you going and, you know, then maybe generate a list of events in sites that you may want to see in a potential trip itinerary and hotels, you might want to stay at, you know, things like that, is that so it's something like that possible? You know, we're looking into that, you know, could it be used for responses to FOIA requests, to be able to ease the searches for that information, and be able to do it in more of a normal language format, for the folks that are working on that process? So lots of different use cases, you know, in the contact center as well. But like I said, we're all just trying to do some POCs. And then we'll step back and evaluate from there.
Alex Whitaker 08:01
Yeah, that's really interesting. So I have a two part question for you. And we've touched on this a little bit. And as you're talking about things that you're sort of talking with agencies about, but I'm wondering if you can tell me first, what is the biggest challenge facing state it that generative it can assist with? And then, you know, kind of related to that? So what are you sort of most excited about looking down the road? What are what are some of the things you think it can really do and and some of the positives that it can bring to state government?
Kevin Gilbertson 08:28
Yeah, no, that's great. The biggest challenge, I would say, is the workforce shortages that we've been having, you know, we're losing folks with the private sector, you know, we've been having high vacancies. And the reality is, is that, you know, that's not going to be something that goes away. Yes, there will be different states and different economies. But at the same time, I don't ever see a situation where we have too many people and too much money to do the job that we need to do. And so I really want to be able to, to capitalize this to be able to get by with less. And what I tell folks, my goal is, is that if I could save everybody, one minute, 30 times a day, you know, over the over the 1000s of employees that we have, that would be a game changer for our ability to provide a better experience for the citizen. And in terms of, oh, I'm most excited about? Yeah, so you know, I think that it goes right in line with that, you know, it's the ability to pick up the pace of what we're doing to move a little bit quicker. I do use Gen AI to some extent, every single day to you know, basically just get an idea. I don't, it's great to not have to look at the blank page. And then, you know, of course, once we start to be able to compile all the mass data that we have in the states, to be able to start using this technology to mine that to develop new systems, develop better processes for working with citizens and providing a better experience--I think there's a lot exciting that come there, that we're going to get more productive, and we're gonna have, you know, better ways to do it.
Amy Glasscock 10:03
Yeah, that that brings up a good point. So, you know, when we started talking about using it to mine state data and stuff, of course, there's a lot of personally identifiable information and sensitive data. And so, you know, along those lines, obviously, we are going to be looking for generative AI tools that protect that data from going out into the world. So you know, what do you what do you think's the biggest threat to state government from generative AI?
Kevin Gilbertson 10:29
Oh, there's, there's a number of threats that we need to be, you know, thinking of it, number one would be simply not taking advantage of it at all, you know, we can't ignore it. And so if we were to do that, that would be I think, just as bad as you know, taking cautious approach to getting here, as you mentioned, you know, sensitive data getting out there, inadvertently by an employee, if that was used a training model or something like that, that would be good. You know, I think of the possibility of spread misinformation fraud, you know, when when you take seconds to listen to someone's voice, and then you can basically sound like them, if you want to, you know, the signs that we use the tech fishing in our security training that we do include things like misspellings, and you know, other things to look for that are going to get, you know, harder and harder to spot. And so, and then, of course, you've got, you know, the legal and ethical risks associated with it. You know, right now, we're seeing lawsuits against these companies for copyrighted information and IP. And you know, what, of that may be passed on to consumers of it. And so, you know, those are certainly things that we need to look into. So sorry, not not a single threat, but there's definitely a lot that needs to be taken into consideration.
Amy Glasscock 11:44
Yeah, that's fair. So, you know, we were just together for the CIO confidential meeting. And I know you've been in a lot of conversations with other state CIOs around this topic. They were confidential, of course, so not to name any names. [Laughing] But do you think do you think the state CIO community has similar thoughts and views on generative AI? And if so, kinda like, what do you think they are? Or do you think it's been a little more divisive for people coming at it from different points of view?
Kevin Gilbertson 12:12
Well, I would say there's folks coming in at different points of view, but I don't know that I characterize that as divisive. But I would say it's just a matter of, you know, the speed at which we want to approach things, the order in terms of what we want to do in terms of the guidance before people start using versus after, and all of that, I think there is a lot of optimism out there, though, in terms of what can be done. Now that said, you know, when we get together, you know, we talked about, you know, what are those threats? What are the concerns? So we've talked about things like safe usage, you know, making sure employees know not to enter, you know, PII or HIPAA data or something like that in there. What do we need to have in our contracts for our vendors, because our vendors are racing to get generative AI into their systems. And, you know, this is not a choice for them. I've talked to several different vendors that are, you know, that are on this race to find, as many use cases as they can, can get to use, you know, use this, and of course, they want it to be safe, you know, workforce concerns, what is this going to do? You know, people are going to be worried about being replaced? And or, you know, are we going to run into situations of burnout, where we are expectations of what somebody can do rise so much because we anticipate that they're using this tool, and so the pace of work will will rise beyond what it probably should. And then you've got the content errs hallucinations, bias. And then, you know, just misuse in general. So, you know, we spent a lot of time talking about that in terms of, you know, how do we prevent, you know, these concerns, enable safe usage, you know, those types of things, I think, are generally top of mind at this point. But I do think that we're all looking for opportunities to take advantage.
Alex Whitaker 13:58
Yeah, that's, that's really interesting. And I think kind of as we look for these opportunities, it seems like the use of AI is really starting to kind of become ingrained and state policy. So I think I know the answer to this next question, but I really want to hear it from you, though. And that is, do we think that generative AI is a fad? Or is it really a turning point in how we use technology?
Kevin Gilbertson 14:18
Yeah, right now, I, I think it's a turning point without question. You know, we've certainly talked about the ethical concerns that need to be taken to consideration. So a lot needs to be done, but a lot of the large organizations that are providing AI tools are coming together to to provide more ethical guidelines to try and figure out ways for safe usage and things like that. So I think that there is some responsibility being taken by you know, those those vendors. So I do think that this will be easier and easier to to use and justify in terms of use, but I really don't see this going away. Because there's so much opportunity here. You know, like I said, there's lots of different use cases in the state but at the end, it will allow the citizen to interact with us in a much better way.
Alex Whitaker 15:05
Yeah, awesome. Thank you. I mean, it certainly does not seem like a genie that we can put back in the bottle. So we're glad that you and your colleagues that are kind of leading the way to figure out how to use it appropriately. So. All right. Well, thank you so much, Kevin, we really appreciate you sharing your thoughts on this topic. But we can't let you go just yet. Because on NASCIO voices, we recognize that we are all more than just our work. So we'd like to ask you three fun questions about your life outside of work, and a segment that we call [thunder sound effect] the lightning round. Are you ready?
Kevin Gilbertson 15:33
I'm ready.
Amy Glasscock 15:36
Okay, so speaking of robots, and helping us with our tasks, if you had a robot at home, what's the one task you would like them to handle for you?
Kevin Gilbertson 15:45
Well, if the outdoor counts, I'd definitely like them to take care of my landscaping for me. Absolutely. Big thing, but, but you could throw in a lot of things, you know, laundry dishes, I already have one that does the floor. So I'm okay there.
Amy Glasscock 15:58
Yeah me too!
Alex Whitaker 15:59
Laundry was gonna be mine. I've got two kids, I feel like I'm always doing laundry.
Amy Glasscock 16:04
Mine would be cooking, but that's pretty advanced robot. [Laughing]
Alex Whitaker 16:06
Yeah. So what career did you hope to have when you were a kid?
Kevin Gilbertson 16:12
I actually wanted to be an astronaut. And, you know, I was a big science fiction reader since probably about the third grade and just thought I wanted to go to space. And then, you know, things got in the way of being able to pursue that dream. But so I never imagined getting where I am here. But I'm, you know, also very happy that I did.
Amy Glasscock 16:31
All right. Well, that ties in perfectly to my last question. So I'm sure you must have an answer. Star Wars or Star Trek?
Kevin Gilbertson 16:39
Well, I think I'm gonna ostracize someone no matter which direction. So I would have to say Star Trek. I've been watching for a long time, and probably some of the older ones that haven't kept up lately, because there's, there's too much to keep up on.
Alex Whitaker 16:54
All right, just you got to get the new ones are great. I gotta say all the new Picard and everything. So check it out.
Kevin Gilbertson 17:01
I definitely will.
Amy Glasscock 17:02
I was raised on a lot of Next Generation, I must say. Yeah. Although my parents also like Star Wars. So I don't know. I guess I can speak with languages and I married a Star Wars person.
Alex Whitaker 17:15
You can speak Wookie and Klingon? Ugh that was so nerdy, can you edit that out please?
Amy Glasscock 17:20
[Laughing] This is the nerdiest podcast we've ever done so far [laughing]. Alright Kevin. Well, thanks so much for taking the time to talk with us today. And also for helping out with this new generative AI working group. I know that you're super busy, but we do appreciate your time for NASCIO.
Kevin Gilbertson 17:36
Thank you. And I look forward to continuing and hoping to get a lot of participation from the community in the work group that we've got put together. I think we can do some some good work together. Yeah. Thank you.
Alex Whitaker 17:49
Thanks again for listening to NASCIO Voices. NASCIO Voices is a production of the National Association of State Chief Information Officers or NASCIO.
Amy Glasscock 17:56
If you liked this episode, please consider reviewing it and Apple podcasts. It helps listeners like you find us.
Alex Whitaker 18:02
We'll be back in two weeks more great state IT content.