Evolving the Enterprise

Harnessing AI in Government: Opportunities, Challenges, and the Workforce of Tomorrow with Adam Leonard, Chief Analytics Officer & Data Evangelist at the Texas Workforce Commission

• SnapLogic • Season 3 • Episode 1

Share episode

In an era where technology and AI are reshaping our understanding of employment, society, and governance, Dayle Hall engages in a conversation with Adam Leonard, Chief Analytics Officer & Data Evangelist at the Texas Workforce Commission. Adam provides a rare peek behind the curtains of government operations, revealing how AI and automation are already driving significant changes in efficiency and service delivery.

Starting with the nuts and bolts of government transformation, Adam shares anecdotes about AI-powered chatbots, their potential to streamline operations, and the broader vision of enabling government employees to tackle high-value, creative tasks. However, as with all powerful tools, AI presents both opportunities and challenges. Adam doesn't shy away from addressing the elephant in the room: the ethical quandaries surrounding AI, from its role in disinformation campaigns and the creation of deep fakes, to the broader implications for the workforce.

Central to this episode is a thought-provoking discussion on the future of jobs in an AI-dominated landscape. Adam posits that the pace of job transformation will likely be unprecedented, leading to an urgent need for adaptive education and training systems. The conventional models of hiring, based on credentials and job titles, are being disrupted, giving way to a more skills-centric approach. How can we identify and respond to emerging skill demands in real-time? Adam proposes innovative solutions, like mining data from job postings, to bridge the gap between emerging job requirements and education.

Beyond the pragmatics of technology and employment, there's a palpable undercurrent of concern for societal well-being. Both Dayle and Adam grapple with the question: How can we harness the immense power of AI without leaving swathes of society behind?

As AI becomes an inextricable part of our world, this episode serves as a comprehensive guide, combining philosophical musings with actionable insights. Dive in to understand the future trajectory of AI in governance, its societal implications, and the proactive steps we can take to navigate the challenges ahead.

Sponsor
The Evolving the Enterprise podcast is brought to you by SnapLogic, the world’s first generative integration platform for faster, easier digital transformation. Whether you are automating business processes, democratizing data, or delivering digital products and services, SnapLogic enables you to simplify your technology stack and take your enterprise further. Join the Generative Integration movement at snaplogic.com.

Additional Resources

Harnessing AI in Government: Opportunities, Challenges, and the Workforce of Tomorrow

Dayle Hall
Hi, and welcome to our latest podcast. I'm Dayle Hall, CMO of SnapLogic. As usual, we've been talking about this for a while now, you should be used to it, we're trying to give all the organizations’ IT and data professionals out there some insights and best practices on how to integrate, automate, and potentially, hopefully, in a positive way, transform your enterprise.

We have a special guest today. And I say special because it's slightly different, more unique than some of our previous guests. Joining us today is Adam Leonard. He's a national leader in utilizing data to enhance governmental program effectiveness, particularly as it relates to education and workforce development. As I said, very unique, we haven't had someone like Adam on here before.

With his role in analytics, evaluation, and business transformation at the Texas Workforce Commission, he collaborates with all the stakeholders to advance data-driven decision-making, strategic planning, and data and program integrity. Adam, thanks for joining us today.

Adam Leonard
Well, thanks very much for having me. I always look for an opportunity to talk about these things. On LinkedIn, I refer to myself as chief analytics officer and data evangelist. So I'm a true believer.

Dayle Hall
I like the data evangelist title. Well, I think it will be good- again, I haven't heard of this type of role. Obviously, it's within Texas. Why don't we start with this, give me a little bit of your background, how you came into this role, your experience, your career and so on, and then talk to me a little bit about what your main focus is at the Workforce Commission.

Adam Leonard
Okay. That's a bit of a story because what people often do when they learn about me or meet me is, or read my resume, they go, so you have a BA in government and a master's in public affairs, and you're a chief analytics officer, how exactly does that happen? And really, it plays into the whole idea of aptitude and interest. If you think about the transformation that has been haltingly underway in hiring around skills-based hiring, and not credential-based hiring, or job title-based hiring, this idea that you're looking for people who can do the work and who have interest in the work, and I just had a natural affinity with data.

I joined the Texas Workforce Commission about 20 years ago in a smaller role, in a smaller department that only focused on official compliance reporting, essentially, the stuff that we have to report to our state government and to the federal government because they give us money to do certain things. And so with that comes performance measures. Some of those measures were good, and some of those measures were garbage. But we had to report them all the same anyway.
Over time, I began to realize that this was a waste of resources. We're spending 80% of our time replicating data, ad hoc-ing data that should have been automated. And so what that meant was we were answering a whole bunch of who, what, when, where type questions, very conditional statements, and not answering questions about why and how and what's going to happen next and what should we do about it.

And so I began to, within my organization, start seeding these ideas that these are the questions that we really should be spending our time on, and was able to get some investment in automation to allow us to take so much of what we were doing in the compliance realm and automate it so that that would allow me to free up my staff to spend more time on these more high value-add questions and let people answer more of the straightforward conditional questions themselves.

And then after that, I began to make the case that if you really want to do analytics, if you really want data-driven decision-making, you can't bring data in after you've done the initiative. You have to bring us in while you're planning your initiatives so that we can provide input on what good looks like, on what data you're going to need to capture in order for us to do an effective evaluation, possibly program design, those sorts of things.

Ultimately, my executive director at the time bought off in this idea. It was a knock on your door, you got a minute unscheduled meeting, and he said, sure, come on in. And I laid out this idea for a division that we called operational insight in about five minutes, and he just signed off on it right there, said, go off and write it up, and we're going to make it happen. A couple months later, I had a division and was on the exec team, and we've been running gangbusters ever since.

Dayle Hall
That must have been some pitch.

Adam Leonard
It was pretty good.

Dayle Hall
It must have been. Well, I know it's interesting, the things you talk about. Things you mentioned specifically there that resonate with me and actually that I've heard from, not necessarily a government kind of entity, but approaching the team that run the data, that manage the data early in the process. That doesn't happen everywhere. But I hear that more from people like yourself, which is when you're trying to solve a problem, don't then start gathering the data. Talk to the team that are more involved in the data because they'll help you. So I think it's very typical that organizations tend to start on a project without necessarily bringing in that data analytics team.

Adam Leonard
I think the thing that helped make this happen is that I took my time in working within the division. I established the relationships. I established trust. And that people understood that I came from a program background. I did 10 years in a different agency, more on the program side as a data consumer, not a data provider. And so I can kind of bridge both worlds and speak both languages. And in doing that, I was able to show them more about the possibilities of what we could do with this data that they were just not taking full advantage of.

So by not being in a hurry and by slowly advancing this agenda and this idea, it got them more ready so that when the proposal actually hit the table and it was announced, everyone I think saw exactly- well, of course, that makes total sense because we were building towards it unofficially. We were already thinking about what can we do better to work together, how do we get this input up front to make ourselves more effective and then use that data as a feedback loop.

Dayle Hall
You mentioned spending a lot of time running reports, reporting on reports, whether it's 80% or 70%. One of the things that I've come into contact during these podcasts with customers, or just being in the market, is that with all the automation discussions and using analytics, and now accelerated with AI, there's this impression that it will take people's jobs, that if you automate things or you use AI, it will take people's jobs.

But I think what I hear from other people on these podcasts and from people like yourself is what it should do is remove some of that burden, the monotony of saying, I'm going to spend 80% of my time just keep running reports because I have to, and they can get more creative. They can do better things with the data because- I don't want to call it grunt work, but some of the stuff that can be automated, doesn't mean that person has no job, it means how about we can use them to be more creative and where the data goes and working with different organizations. So that was a thing that I think you said up front. But that is a mind shift, I think, of people.  Do you see it in your role?

Adam Leonard
Oh, absolutely. One of my jobs is to try to automate out the mundane. The reality is that standard work is something a computer should do. Humans are more likely to make mistakes doing that. So let the computer do that. Let humans be creative. Let me invest more in my staff, and let them do higher-value work. Let me pay them more money because they have higher skills now. Let them have greater satisfaction because they're not just doing the exact same thing over and over again. And most importantly, let's improve morale amongst the staff, at least within my group say, we're not doing this because we have to. We're doing this because it makes people's lives better.

That's what government is supposed to be doing, is we're supposed to be- like my agency, it's about helping employers and individuals and families and communities in Texas. So if we make our programs better because we're able to find things that work better for different types of people through the data analysis, then that means we're changing people's lives. There are a lot of people who want their work to actually mean something. There are a lot of people who chase the dollars with this skill set. And Google and Meta and the big companies in Austin are going to probably hire them.

And then there are others who are looking for some balance in there. It's like, yeah, of course, they want to make decent money, and we can pay decent money. We can't pay crazy money because we're government. But we can give you purpose. We can give you opportunity. We can give you a fast track to development. We can invest in you. And we can even go to bed at night knowing that your work is changing people's lives.

Dayle Hall
Yeah. And as you said, I think if you can take out some of the mundane, repetitive work or effort, then also they'll feel more fulfilled. Again, if someone wants to go and work in one of those large organizations you mentioned, then that's fine. You make it a better environment where you are because they're doing meaningful work and because you've removed some of that mundane.

Adam Leonard
And if I to help a person, a new, fresh out of school person, no experience, become experienced, they have improved their skills, and they go to work for a private company in Texas, haven't I helped an employer in Texas.

Dayle Hall
Absolutely.

Adam Leonard
I mean, that's part of the mission, too.

Dayle Hall
Absolutely. Well, I have a bunch of questions, different sections. I want to start with something that is just more of a baseline because I think within your organization, and specifically the work that you do, it's slightly different to some of the enterprise corporations I talked to. When I reference the term, and you talked about this in a number of your public posts, when we talk about democratizing data, how would you define that within the environment that you work in? What is democratizing data across a state local government?

Adam Leonard
I would say this applies more broadly- a version of this can even apply in the enterprise, which is that, to start with, the people who do the work are in the best position to have an impact. But they need to be armed with good information. They need to be able to answer questions on their own without having to know SQL, without having to know Python. They need tools that maybe they could do some stuff in Excel, but the reality is they just need access to good data that they can work with.

And so by democratizing the data, by putting that data out there so that they can not only just run canned reports but actually do a certain amount of their own custom filtering and conditionals, then we're spreading the footprint of exploration. And then when one of them finds something interesting in the data, because they've done some custom set of filtering, and they don't know where to take, they don't know- they found something, they don't know what it is, they don't know why it is, then they can come back into my division and say, look at this, can you explain this? My folks might not have ever found that. There's too much data. We've got over, I don't know, 25 programs as an agency. We're a huge agency. But we've got all these people who are operating the programs. And if they can look at the data and work with the data and apply the data and ask questions with it, then we're better positioned to try to better utilize it.

Within my division is also our business transformation function. So think about Lean Six Sigma Theory of Constraints, those types of things. And so that's a big piece of this also is about using data to help improve processes. Our premise in all of that work is the idea that the people who do the work are in the best position to improve the work. But even they- again, the data, they may not realize- they know something's not working, but they can't figure out why because they don't have the data to figure it out where the bottlenecks are, where quality is falling off, etc., where the mistakes are being made. And so there's a play in there.

But for governments, perhaps different than private enterprise, democratizing data is also about making the data publicly available, where stakeholders, partners, other agencies, and even the true public, is able to work with tools and see on their own, answer questions on their own, many of the same things. So think about in terms of transparency, open government, those kinds of things. That's good for accountability and trust. I'm sure it's the same everywhere.

But people love to make jokes about government and how far behind we are and close enough for government work and those things. And yeah, there are organizations like that, that have earned that kind of reputation. But the way that you change it is by showing them that, no, we're actually doing something that's not far behind where the private sector is in terms of utilizing their data to improve their mission.

Dayle Hall
Yeah. And I think you hit the nail on the head, which is I think there's a level of scrutiny under government, federal, local, whatever. There's a level of scrutiny because of that public nature of the data that they have to put out. I really liked what you said, which is the people that do the work are going to have the biggest impact if you give them the data. And I think that is a key, whether it's enterprise or government, if the people that really understand a function or a process or whatever, if you can help them by giving them timely access to the data, they will be more successful.

But on the government’s side specifically, when you talk about democratizing data, and there is that public nature, does that make it more of a risk to have access to this type of data? Do you have to have more scrutiny? Do you need more checks and balances? Is there more of a risk with democratizing data because of that public piece?

Adam Leonard
Well, there's definitely a greater risk of embarrassment because you're putting it all out there. But a lot of it, the other risk, if you will, like privacy, to personally identifiable information, to disclosure, is a matter of how you build it. The way that we're building our systems is that they all have- we're working with Tableau is our primary data exploration and visualization tool. We've developed code that automatically suppresses cells. If the cell size gets less than x, and x is usually 5, it just puts an asterisk there, and you don't know whether it's 0, 1, 2, 3,  or 4.

And the reason for that is because if you select down far enough, I know county, age range, demographic information, certain things like that, education level, I might, could get down to a level where I have a reasonably high shot of identifying the person in the data because only one or two people show up, and I can guess that right.

And so if you think about from the standpoint of people who are interested in identity theft or who are interested in- I don't know, let's just say they had a poor relationship result, and they haven't given up on it, and they're behaving inappropriately towards another person. Yeah, those are things we really are concerned about and why we make sure that we have these privacy enhancing tools automatically built into the things that we're publishing. And there's no ability to get to the backend data. There's no ability to download that data.

We're also starting to experiment with, or at least look into experimenting with, might be a better way of putting it, synthetic data so that we might can make versions of data, grab more granular data available that don't represent real people, that could continue to maintain the statistical characteristics of the real data so that it's useful for researchers.

Dayle Hall
Yeah. Well, that's interesting. Is it still the same skill set around the professionals that work in and around the government area? Are they still thinking about data to drive decision-making processes around business operations? Are they looking to solve more of the challenges within the environment, within Texas itself? I'm trying to think, where do they start with data? Are they trying to solve more of a business process and have access to that? Or do they get access to the bigger picture because of the public data?

Adam Leonard
The public is probably more interested in the program outcome-type data. So they want to know more information about, okay, how many people are you serving, what types of people are youserving, what types of outcomes are they achieving in terms of employment and earnings and retaining employment, achieving credentials. They are interested in the degree to which the pool of people we're serving is reflective of the overall population, so think demographically compared to census data, looking to see whether there are differences in outcomes by different ethnic or racial or gender or age categories that are atypical. So there is interest in that.

And quite frankly, we have interest in that too because if we see things in there, whether they might be biased or they just might be- that program is not as effective in serving certain populations as others, then that's something we want to try to improve upon. So I think that the public is more concerned about those things than they are in knowing how long it takes us to execute an RFO. That's a business process, and we track data associated with it. But that's not really a priority to publish that information when we've got so much other data that probably can have a greater direct impact in our system first.

Dayle Hall
Some of the things that I hear from people that I talk to, customers, prospects, is their guidance is always, look, start with a business outcome, what are you trying to solve, and then work with the data team to make sure that are we capturing it, where are we capturing it, how do we use it, and so on. In your scenario, do you start with the data you're meant to provide to the public first, or do you start with still what is our process, what is our business process we have to go through to capture that data? What do you typically start with?

Adam Leonard
We typically start with what's the data we're required to capture under federal or whatever regulation because that's the minimum cost of entry. I'm sure you wouldn't be surprised to learn that there's a lot of data we're required to capture for most of our programs under federal law or state laws, more so under federal laws.

And so there's a lot of data there. Demographic data, there's also a lot of data about what different types of factors that might affect somebody's ability to obtain and retain employment. Perhaps they've had run-ins in the criminal justice systems. Perhaps they're from a low-income background, they didn't complete high school, things like that. We have information around all of that that helps us both to analyze the data but also to help understand the specific challenges that a person has so that we can try to work with them to minimize the impact that those things have on their personal lives. So the more trust there is, the more willingness there is to share between the person that we're serving and the people who are serving them, the greater the likelihood that we can bring the right resources and services to bear to help them.

But taking it up a level, something that I've started doing more recently, we're currently building out our next-generation analytics infrastructure environment, and we're starting with the compliance stuff because we've got to do the compliance stuff. But as soon as we finish all of that- and we are making some enhancements beyond compliance at the same time, but as soon as we finish all of that, my approach to requests for, hey, I need a dashboard is going to be, so tell me about your business process. Show me how it works. Show me what the dashboard is intended to support. Tell me how the dashboard is going to inform actions on your part. Like when it goes above or below this control point, what are you going to do? So that way, we can be sure that we're building you a tool that's actually going to inform your process and help you make decisions.

If you don't know what you're going to do, then that's not a dashboard. That's an Excel project. We just need to ad hoc some data for you to look at, and then you can figure out what's meaningful or not. Then you come back to us, and we can talk about automating it and building you a tool to manage with. But it really needs to be thought through that way.

One of my favorite expressions or things that I tell people who are involved in data work, especially people who are my peers, is the surest way to do work twice is to just give people exactly what they asked for the first time without questioning them, without making sure you really understand and everything, because they always come back with, oh, what I really meant or what I really needed was this other thing.

Dayle Hall
Yeah. I have a question. You said something around- because you start with the compliance and the data you have to provide. Does that slow things down? Because I think about healthcare, financial services, government, there's more rules, there's more regulations. You mentioned it yourself around there are sensitive data, there can be bias in the data. Does that make the job that you do and your teams, does that make it harder because of the compliance? Or is it just a matter of figuring out the process and then making sure that you keep that in the back of your mind? I just wonder whether a lot of the enterprises- like if I'm a marketer of a software company, which I am, but I'm not necessarily dealing with as sensitive information, so I can probably- maybe I can move quicker to capture it. Do you feel like it slows your process down?

Adam Leonard
Yeah, no, it can. In my agency, I'm a direct report to our executive director. I'm totally outside of IT. My division is a hybrid of technical and business in nature in what we do. While my group partners with business and we partner with our IT division in terms of the data that needs to be captured in the transactional systems, whether from compliance standpoint or for something that we think we need in that business supports are getting, the reality is that when Congress gives you a few hundred million dollars to do a thing and says you're going to have to report this stuff as a result, that's the price of admission. Those are the strings that come with it.

And so just like I did when I first got here and said, okay, we have to do this, but do we have to spend all of our resources on it, same thing now. Okay, we have to do this, we have to make it better than it was, but let's figure out a way that once we rebuild it, that it's going to be something that we can easily modify and grow with, and it's going to answer more than the minimal questions that were in the more than the minimal compliance stuff that we need, and we can go further from there.

Dayle Hall
Yeah. And I really liked what you said about if you want to duplicate the work or have to do it again, give someone exactly what they ask for without necessarily going in and questioning it. In your environment, do you find that you get a lot of requests for data without that kind of like, okay, what problem you're trying to solve,, outside of the compliance stuff?

Adam Leonard
Well, we used to. And we've changed the culture in the division to say, you gotta go back and clarify. You have to really understand what this is. And we've explained to the business users why we're doing this, that look, this is in service of meeting your need the first time the right way. We're not trying to be difficult. We really want you to have the thing you need. It's just there are some times there are people in the business areas who know enough about the data that they'll get real specific about what fields and tables and things and not realize that's like, well, there's a better table, or an element doesn't quite work the way they think it does and there are better choices to use that would get them exactly what they're looking for. So it really does get down to what problem you're trying to solve and let us help you do that.

Dayle Hall
And that comes back to what you mentioned, which is bringing your kind of function earlier in the planning because that will probably save multiple iterations for you, and it will also give them probably better data and the data that they look for the first time.

Adam Leonard
Yeah. And years ago, before we got better at communicating and coming in early, it used to be they’d come to us, say, hey, we did this initiative and we're ready to- we need to know how it works. And we're like, okay, what data did you get? And then they say, well, just the normal stuff. And then we looked and looked what their initiative was, and it was like, we can't do anything with this. And so there were a lot of good ideas that never really got properly tested and evaluated back then that now get much more analysis and focus.

Dayle Hall
Yeah. You mentioned something a couple of minutes ago, the word culture. I want to talk about  that a little bit. In terms of culture, what do you define as and how do you think about establishing a culture that promotes, I don't know, if you want to call it data literacy or promotes this decentralized concept of data? Where do you start with that?

Adam Leonard
Culture is a big deal to me. The reality is that it's not enough to have a data evangelist like me. I can do a lot, and I can be very persuasive and try to get people caught up in it. But the reality is that for this to be effective, your CEO, or executive director, or commissioner, or whoever is ahead of the agency, has got to be a true believer. They can't just mouth the words. They got to say, this is what we need to take our system, our agency, to the next level. Our data represents our past and it informs our present. And if we're smart about it, it can guide and predict or even help change our future. That's the message that we have to get out. And I can talk about all this stuff. You can listen to me do it for 30 minutes. But at the end of it, if it's coming from me and it's coming from our boss, then maybe it's going to get through a little more.

But the other thing is it just takes time. You have to have patience. There was a- many, many, many years ago in Texas, I've been in Texas government 30 years. And towards the beginning of that experience, they wanted to do total quality management in Texas. The governor was all excited about it, wanted to do it, and asked them, all right, so how long is this going to take to put in place? They said 20 years. And she was like, no, really? And they're like, yeah, no, it takes 20 years to do this. I don't think it takes 20 years to get to a data culture. But the point is culture doesn't happen overnight. Some of it happens because of repetition. Some of it happens because you're demonstrating with quick wins and starting to win some people over who may have been a little skeptical. Some of it happens due to turnover, that skeptical people move on and others come in who are more ready for it. It is a question of helping it permeate throughout.

One of the things that we also do is we try to talk about the work. We talk about the different things that we're doing. We try to get our partners excited about new products that we're building, showing them what they can do with them and supporting them on it instead of saying, here you go. I think there's a mistake that a lot of data shops make, which is they subscribe to the “if you build it, they will come” theory of data product design. That is not a strategy. That's a prayer. Maybe someone will come and maybe they'll know what to do with it. That's not a really good strategy to success.

The strategy to success and to building that culture is through engagement, training, support, reinforcement, taking feedback, making the change, and when you make the change, you credit who gave it to you. You say, well, so and so in this division suggested that we make this change, which we've done here, as you can see, and as a result of this great idea, now you can do this other thing that we didn't think of. And that gets them buy-in. And once they start seeing that you're listening and partnering,and that what they, the people who do the work, what they think matters, that goes a long way towards the development of culture.

Dayle Hall
It's a small nuance, but I've never heard anyone describe it that way, which is when someone helps you get something else in the data or gives you something, actually, you give them the credit for it. They asked the right question or they looked at it in a different way. And I think I haven't heard anyone really refer to that. But if you do want to change culture, giving people recognition and credit for the work that they do, not just what the final report looks like but some of the nitty-gritty data, I mean, I think that could be key in terms of driving the right behavior.

Adam Leonard
It’s also good for ownership.

Dayle Hall
Right. I have a question around, whether it's CEO, or executive director, or commissioner, like you said, if they're not as progressive in thinking about how to do this, or they need to be convinced a little bit more, what do you think works when you sit with that leader of a business, a function, a group? How do you get them to be that supportive? Because you said, if it's not just you, the data evangelist, if it's your executive leader too, that carries more weight. So how do you make sure that they're on board with it?

Adam Leonard
Yeah, it does carry more weight because they are empowering you and telling everybody's like, listen, I'm a believer in this. This is something we're going to do, and you really need to listen. We did the same thing with business transformation, which was originally its own little division. So it's the same idea.

But I guess to answer your question, because there are probably a lot of people out there who would love to make this kind of pivot in their organization but are facing some resistance, is I would bet you that in every organization, there is a leader of a division, a department, a group that is dying to do something with data. They just know that there's something in the data that would make their group more effective, but maybe they don't have the skills to do it themselves. You find that person. You partner with them. You get quick win. You publicize that quick win. And they talk about it as well. It's a partnership. And you show how you did it together. And you start stacking some of those things. And you measure the impact that they're having. And it becomes hard to ignore. That's really important.

For us, we were already established as a division before the pandemic. We had a great national reputation as data innovators. But if there was any doubt in what we could do as a system working together, it was completely eradicated during the pandemic where we worked with partners in IT, and in the unemployment insurance division, and in our fraud, deterrence, and compliance monitoring division, and all across the agency, people involved in the UI program process to develop an identity theft fraud system to detect that. And we got some outside help to get us started on a few things because I was in a position at the time where I'd been bringing on a bunch of new people with great skills but not a ton of experience yet. We ended up building a pretty much first-in-class type system that since it went live has saved over $4 billion in fraud.

Dayle Hall
Wow. That's a good ROI.

Adam Leonard
Yeah.

Dayle Hall
Well, I think, again, what was interesting, you said there's someone out there that's probably looking to do something better, so find a champion. And the other thing, and I hear this a lot with large projects that are based around a significant amount of data or systems that have to get integrated or automated is it's the quick wins. And it's interesting to know that you look at that too, which is regardless of whether you're an enterprise or a government, it's finding the champions and getting some of those quick wins. So you get that immediate, like, oh, I didn't realize we could do that with the data. I'd never seen this. I'd never looked at it that way. And I think then, as we're talking about culture, the more you publicize that and people see the value, it gives you the remit almost to then go and do the bigger projects because of people's immediate value. Because you mentioned it yourself, a 20-year project, most of us are going to move on by the time that comes to fruition.

Adam Leonard
Right. The other thing is that when you think about analytics work, if it's not life and death that you're working on, the reality is that I'd rather have several projects that meet a moderate causality standard in terms of the study than one that meets a high standard of causality, because it takes so much more work to do a randomized control trial. And I realized, yeah, that's the gold standard. But you know the time and resources are going to take to do that. I can do 3, 4, 5, 6, 10 quasi experimental design studies and start helping a lot more people a lot faster. So to get started with those quick wins, it's like, hey, if you got nothing, good is good. Don't aim for great. Don't aim for excellent. Don't aim for the best yet. Start with good because good is better than nothing. And then go to very good, and work your way up from there.

Dayle Hall
Yeah, no, that's interesting. Just a quick question around managing data because we talked about democratizing it across the organization. But in terms of your role and how you manage it, do you have a model which is controlled centrally? Is it decentralized in different functions, and then there's a way to bring it together? Or is it a mix of both? What's your sense around what works and what doesn't?

Adam Leonard
I think that our organization is going a bit of a transition because we have been replacing a lot of our main transactional systems, our main case management systems, and in some cases, it's been by modifying the existing homebuilt system. And in other cases, it's been buying a vendor product and standing that up. And so that becomes a little challenging from a data management standpoint because while the vendor product, they're in charge of maintaining and enhancing and making controls a lot of times- because if you're going commercial off the shelf, the more you customize that thing, the less it's going to work because they're sporting a bunch of other states. And so then they gotta go back and do all your custom changes all over again. So it's very tricky.

Within our world, what we've basically done is we have an enterprise data warehouse, not a data lake, not anything like that right now. Data lakes were only just really starting to come into being at the time that we launched the data warehouse, and we had people who were really good with Oracle and that sort of thing. So that was the direction we wanted at the time. We get data out of all of these different systems into- it's a data vault structure. So we get copies of the data from all these systems every day in a raw vault format. So it's just a snapshot of the existing system. That then goes into a data vault table where it's looking for changes since the prior snapshot so that we can see how the data is changing day to day. And then we build my division. IT does all that. And then once it's there, my division takes the data and transforms it into business vaults that answer different types of questions. That's just simplifying and standardizing the data for the users.

We have a centralized governance body, steering committee, that is made up of business and IT and my group. There are certain people who tend to have more of a role in the process just because we're closer to the data in terms of working with it and everything. But the owners of the data, IT does not own data. I do not own data. The program areas, they own the data. It's their program, it's their data. IT is in the business of helping capture that data and helping ensure its integrity. I'm in the business of helping transform that data into something useful for them to monitor their programs, make improvements, and make decisions.

Dayle Hall
You find that because the program areas, because they know they own the data and you're there to help facilitate, get more information and insights, does that make them more of a partner rather than saying, well, we're the central function, we own all the data, and you just tell us what you need? Do you feel like that promotes partnership?

Adam Leonard
That wasn't my intent. I'm a true believer that the- I'm a support function. My group may exist because the law requires somebody to do this work. But that's not why the law exists. That's not why the programs exist. The programs exist and the law exists to serve people. And so that's why it's the program's data, because they're the ones who are charged with doing the work and ensuring that the right information has been captured and brought in.  We're there to help them. We are a force multiplier.

And so in doing so, in positioning ourselves as partners, hopefully, that makes us more effective in working with them so that they don't see us as having our own agenda per se. I have an agenda at times, but it is to advance this vision of data in service of others. But again, it's always about the organization's mission and each division’s individual piece of that mission.

Dayle Hall
Yeah, I like that. And as we come towards the end of the podcast, there’s one one topic I definitely want to cover with you. I see it in the top left corner of your background, this concept of data for prosperity. Could you elaborate on that concept? What is it and how is it relevant within what government are trying to achieve?

Adam Leonard
This is my personal movement, if you will, or at least I've kicked it off. It's like fetch, I'm trying to make fetch a thing. I work mostly in the education and workforce development spaces, which means it's about helping people get jobs, helping people retain employment, put people on paths to earn more money over time to be able to support a family, make good decisions about their education and training opportunities. And if you think about it, all that is about helping people achieve prosperity. And my agency's mission is explicitly laid out that way. It's more wordy, but I'm going to boil it down.

Our job is to help employers, individuals, families, and communities achieve and sustain prosperity. My mission as a data shop is to use data in support of prosperity. I think that that is not just within education and workforce. I think that that is an essential function of government in general, that government- if you go back to Adam Smith, government's job is to protect from dangers outside the borders, within the borders, and then to do those things which the private sector cannot do, to help ensure prosperity. The word prosperity doesn't appear in the Constitution, but it does talk about the general welfare of the populace. So it's all in there. It's in Federalist. It's why government exists.

As far as I'm concerned, I am not a policymaker, I'm not an elected official, I'm not an appointed official. It's not my job to decide what government should do. It's not my job to decide how much we should spend. And it's not my job to decide how the money should be raised. That is absolutely the province of elected, appointed officials and other policymakers. Ultimately, whatever they decide, that's what will execute. But what I think everyone can agree, regardless of what affiliation they have, is that whatever we choose to do in those first three questions, we should be doing it efficiently, effectively, and having an impact. And so that's what data for prosperity is all about.

Dayle Hall
I like that. Do you have any examples of how you've seen the work that you've done have a positive impact potentially with a certain official or within a certain project or a certain group? Because, again, talking to you is a little bit more unique for this podcast. So I'd love to hear, how does this work in action? How have you seen it have this positive effect?

Adam Leonard
As a movement, this thing's actually relatively new. I coined the term maybe in February and trademarked it under Creative Commons so that anybody can use it. I want it to be a brand that people can get behind. But there are a lot of different efforts underway right now around data and government that fit within this. And I've been involved in a lot of it. I'm not taking credit for it, to be very clear, but I'm trying to support it and leading parts of it or contributing to other parts.

And so one of those has to do with something in the United States called the multi-state data collaboratives. There are three collaboratives right now that are working together on solving common problems. Because reality is states are different, and yet they're not. Every state has the same basic programs, the same basic needs, the same basic problems they're trying to solve, and yet they solve them all individually. So like all the data problems they're trying to solve, every state has the same data problem. But rather than working together to try to come up with a small set of solutions and let the states choose the one that fits their culture and their values and maybe what they can afford, etc., what works for them, all 50 states are working on their own doing their own thing, or in some cases, they're not because they don't have enough money.

But within a collaborative structure, coalition of the willing, if you will, you're able to bring people together around common questions and common problems, and they can learn from each other, share ideas. And in many cases, there are states that are sharing data across state lines. That's especially true in the Midwest data collaborative, where you have a lot of people who live in one state and work in another, like Indiana and Illinois, or where there's a fair amount of cross-state movement, like people, they went to school in one state and they moved into another one, and trying to understand better what's happened to these folks. That's out there. I think that's making a real difference.

During the pandemic, one of the biggest questions everybody had was, well, what the heck is going on with all of these unemployed people? How many of them are returning to work? Which types of people are returning to work? What types of jobs? Well, as part of the Midwest data collaborative, Illinois built an unemployment to reemployment Tableau dashboard that allowed them to time series, show how the pandemic passed and what types of people were going back to work in different parts of the state at different points in time. It was very powerful.

Dayle Hall
Yeah. Is there a possibility that the concept that you created, the data for prosperity, whilst it's very clearly something that federal, state government should care about, what about the private sector? Is that something that other organizations could potentially go on and work with governments for prosperity?

Adam Leonard
Absolutely. There are a lot of vendors in the government technology space, meaning that they build products to support programs. I mentioned how we have procured case management systems for some of our products. There's also a- I worked with the Bill and Melinda Gates Foundation on a few grants. They've been good enough to fund me on some of the ideas that I've had. And they brought me into contact with a group called the P20W+ community of innovation. Now P20W+ is a horrible name, but it's in national use. It's like if you don't work in education and workforce, you have no clue what I'm talking about. What it stands, the P is for preschool, the 20 is for professional, PhD, JD, MD level, the W stands for work. And so there are these longitudinal systems that allow you to evaluate the effectiveness of different programs, different schools, different states, and looking at longitudinal employment and earnings outcomes and things like that.

So this group, this community of innovation, has been set up and supported by the Gates Foundation, and I'm a member of their core team. They, I'm going to say they because I've contributed but I didn't do a lot of heavy lifting on it, I just tried to lift, developed a common kind of reference architecture around what a modern P20W system needs to have in it. That would allow vendors or states who want to build their own products kind of a road map, a blueprint for what they could do, in order to build a modern system or to procure a modern system and recognize what one would look like.

And within that group, there are a number of the big players in education technology. Let's just say some of the large companies that you're probably very familiar with who do work in this space, who have products out there, so companies like LinkedIn and Microsoft and Google and others who are very interested in this kind of stuff. And it's not a vendor community and we're not here to advise them. It's really more of a question of bringing together some of the ideas from the public sector and the private sector to develop generic ideas that can be then shared with everybody.

Dayle Hall
Yeah, I like that concept a lot.  I have one last question for you before we wrap up. There's the hottest topic right now generally in and around technology is this concept of generative AI, i.e., it's been a while, the concept of generative has actually been a while. But now, with some of the LLMs, it's become more mainstream and there's more people that are noticing it, potentially using it. How do you see the impact of generative AI? Or are the things that you're excited about within the realm that you are, within government, that you feel like that could be a help? Or do you also have concerns that it also could be used for evil?

Adam Leonard
God, am I glad you asked me this question. This is absolutely one of my obsessions right now is artificial intelligence in general, generative AI in particular, both from the ethics standpoint of things,so within development and deployment, the potential for misuse in terms of disinformation and deep fakes, that kind of stuff. But also much more practically within my job. I think the jobs are going to change faster than at any time in history as a result of generative AI. And I think that means that we, those of us in education and workforce in particular, are going to have to react to that change faster than we have ever proven we can before, because there are going to be a lot of people whose jobs are affected, and they're going to need training to help them work effectively within an environment where their job is changing.

Now at our agency, our message is very much like the one that we talked about at the beginning of this conversation. We automated a bunch of canned standard stuff to free people up for higher value-add work, more creative work, more interesting work, work we could pay you more money for. That's how we're approaching it at TWC, is that this is going to allow us to get more work done, because the reality is there's always more demand for government services than there are resources to get them done. So if this helps us be more efficient, great. But that doesn't mean that we wouldn't necessarily have to retrain some people to more effectively use those tools. And I think that's going to be a huge challenge for us. And I'm very concerned that nationally, if we don't really study this and put ourselves in a position to reform education and workforce development to react to it, that we get tens of millions of people who are really materially affected by generative AI in a negative way.

I'm not trying to be a doomsayer. I'm not ignoring the projections that there will be a net positive number of jobs. But saying that there's going to be a net positive number of jobs is like the old expression that you can drown in an average of 6 inches of water. Six inches on average, but there's a deep end there, too, right? And so not everybody- even if there's a net number, your job, not everybody's going to be able to do them if we don't prepare ourselves to help them. So we need to understand what they're going through, what the changes are. We need to communicate with employers to understand how they're using this technology and what we can do to help support their workers to be productive for them.

Dayle Hall
Are you having those discussions now internally within your organizations you're in now, which I think- actually, that fills me with confidence that people like yourself are actually thinking about this kind of stuff ahead of time.

Adam Leonard
I'm working on that. I'm getting ready to speak on a panel in two weeks to talk about skills-based hiring and my idea about how we can look for changes in the workplace by mining data through artificial intelligence in online job postings to look for changes in skills that employers are demanding, and feeding that information back into the education and training systems that are out there in order to hopefully shorten the cycle time between signal and response. Because the problem with the historic method, and I think employers have been struggling with this, which is why they're moving away from or trying to move away from traditional hiring methods, is that credential-based hiring, credentials were always just a proxy for skills. They were just a way to try to figure out who to interview. But you still didn't really know whether the person was qualified. You just knew that they had a four-year degree and you think that degree teaches people certain things.

And then you look at job titles and you think, well, job titles, that probably means they know how to do certain things too, right? But skills is a totally different thing. So if we can see specifically how the skills that they're asking for in their postings are changing, that's where the emerging demand change is going to come from. That's what we need. That's the signal we need to be able to detect early enough to feed that information back. Because creating a whole new degree program, that takes years. But changing your curriculum to include two weeks on a new skill, you can do that in a semester.

Dayle Hall
Yeah, that's great. Well, Adam, I appreciate your time. I'm pretty sure we could go on for another couple of hours. But maybe we'll save that for a second follow-up podcast. Adam, I've really enjoyed the discussion. Again, a little bit more unique to some of the other organizations that I've talked to, but I really love how you're thinking about it. And I really appreciate what you said around the work that you and your organization do has a massive impact on society and people's- you mentioned it yourself, it's not just their work, it's not just their employment, their career and so on. It has a broader impact. I'm glad that people like you are thinking about that. So thanks so much for being part of the podcast.

Adam Leonard
I do appreciate it. I'd love to have an opportunity to talk about these things. And if anybody's interested in reading some of my miscellaneous thoughts, I'm at linkedin.com/adam-from-texas.

Dayle Hall
Nice. When this gets published, we'll make sure we promote this far and wide. But in the meantime, Adam, thank you so much for being part of the podcast.

Adam Leonard
Thank you again.

Dayle Hall
Thanks, everyone. That's the end of our episode. We'll see you on the next one.