← Priorities Podcast
Priorities Podcast
Chief AI officer role is ‘part strategy, part operator, part change leader’
00:22:32
Chief AI officer role is ‘part strategy, part ope…
Loading summary
Transcript15 lines
- [00:02]
A
Hello, you're listening to State Scoops Priorities podcast. I'm Colin Wood, State Scoop's editor in chief. This week you'll hear from Ty Fan, who last November was named Oklahoma's chief AI and technology officer. We'll get to that in a moment. But first, here are some of the big headlines in state and local IT news this week. State technology officials last week brought before a House Homeland Security subcommittee a request that Congress reauthorize funding for the expired state and local cybersecurity grant program and renew cybersecurity programs inside the Cybersecurity and Infrastructure Security Agency that have been decommissioned under the Trump administration. Colin Ahern, New York State's Director of Security and Intelligence, said, quote, our states are on the front lines of multiple cyber conflicts, yet we are being asked to manage nation state risks while our federal partners step back the Trump administration's Federal Emergency Management Agency Review Council is proposing one of the most significant restructurings of the federal disaster response agency in decades, a shift that could dramatically expand the responsibilities of state and local emergency management agencies. A report published by the council this month calls for expanded communication systems and new systems for tracking assets across the various levels of government. California Governor Gavin Newsom last week signed an executive order directing agencies to prepare workers, small businesses and the public for the potential economic disruption brought on by artificial intelligence. The order directs agencies to study potential labor market shifts tied to AI adoption, including layoffs, hiring changes and skills gaps. On this week's episode, Tai Phan, Oklahoma's chief AI and technology officer, shares what he thinks are some of the most promising uses of AI in government and his approach for implementing them. I asked him about common misconceptions around AI and the maturity level of AI in his state. His role is a relatively new one, so I began by asking how his role is defined and who he primarily works with.
- [02:13]
B
We were very intentional about designing this role for the state of Oklahoma just because of the trajectory that we're on. You'll notice that there's really two pieces of my role the things before the ampersand and the things after the ampersand. And I think the one that precedes that is really what most states have today, a leader who's really focused on ensuring that the proper strategy safeguards are there to help with speed and execution, as well as scaling capabilities tied to AI and data. I think what's unique about how we're positioning this in the state of Oklahoma is that early on there was a realization that because AI is so infused into all facets and all aspects of, of enterprise technologies, both in terms of people, process and tech. It was very important that we positioned it in concert with some of the other pillars that had a lot of demand as well as impact to citizens. And so in my seat it also encompassed enterprise data management, AI as well as connected experiences. We're calling that connected experience here in Oklahoma. Other states may call digital experience digital strategy. It's really the citizen facing surface of all government service products as well as everything else that we provide. So for us, strategy here means quite a few different things in respect of what our state is and how we're structured. We service about 125 different agencies, boards and commission. And when you roll all that up in a very logical way, you get about seven to 10 different industry domains, healthcare, finance, direct to consumer or retail experiences. And so it's important that we position a strategy that provides both efficiencies and economy of scale for things that are considered statewide enterprise services. So cloud data, application development. But then also the strategy means we have to make sure that we're really ready to deliver domain aligned solutions so that when we're building something for dhs, it has enough reuse for other domains outside, but relevant enough to drive strategic and mission objectives within that organization and that specific vertical. We do a lot with agency leaders, of course, based on what I've just said, but we also work with private sector partners as well as other adjacent ecosystems here at the states, particularly higher education. We're finding a lot of great synergies in their journey in navigating AI as well as ours here in Oklahoma. But also I would say that the one big shift in how we thought about the teamwork and the collaboration moving forward is this is the one sport, if you will, that is that requires everybody to be involved. Legal, procurement, risk, cybersecurity, policymakers. We're increasingly seeing a need to get those leaders involved early on because the decision for AI should be done collectively and it does take ability to drive the transformation. Well, and so this role is no longer a technical role, at least in Oklahoma. It's part strategy, part operator, part change leader. And also I would say the one thing that's been really interesting to see unfold is how do we actually kind of ring true to the promise of what this is from a government standpoint, which is how do you build that trust appropriately? And it's challenging and also fun to do when we're thinking about it in context of AI.
- [05:35]
A
Yeah, you came, you came dangerously close there to saying AI is a Team sport. So you have to, you have to be careful. The cybersecurity people are going to, going to come for you for, for stealing their slogan.
- [05:47]
B
Hey, they probably already know I'm saying it right now.
- [05:50]
A
So. All right, so given all of that really collaborative, cross agency, cross discipline work, where is Oklahoma at right now in terms of maturity level with both AI and digital services?
- [06:05]
B
Our AI journey started officially back in 2023 with the Governor Task Force for AI. So let me give you a little bit of context there. At that time it painted a really amazing depiction of what the future could be. A vision for the art of the possible of AI broader emerging technology. Since our estate IO, Stan Dan Cronin has taken seat over the last year and with the creation of this role over the last few months, we're really transitioning into more of the art of the real. And for us, what we set out to do were a few things that, you know, I'm happy to share now we are moving towards a governed, scalable approach when it comes to AI. No longer running pilots, but really producing hardened capabilities that serves all those domain verticals that I mentioned earlier. And so what we've done recently is we have built AI for AI, an accountable innovation model that really anchors into four key pieces of how do we think about scale and driving efficiencies. We've released a shared enterprise platform, a governance foundation, if you will, that allows agencies to leverage capabilities that are pre reviewed, pre trusted so that they can see the benefit of not starting from scratch, see the benefits of having a shared cost model for producing custom solutions moving forward. And we're calling that Beacon. Another part of our strategy is Lighthouse, where we're still doing pilots, but we're moving concepts into production in a more controlled pathways. Those various different collaborators that I mentioned as part of this team sport for AI concept is really part of that journey. Pilots are not done in isolation. They're done as part of, of a joint review across the board for all those stakeholders that has a stake both from day one to the day that it scales and launches. I would say the third piece around what we're doing that's helping us drive governance at scale is a stronger focus on literacy. And here this is more than just about adoption or getting licenses into the hands of folks. It's really about ensuring that the workforce is ready, our job families are starting to match what the actual fast tool needs. And I think this is a nice moment where we're, we're quickly seeing that an AI transformation has to be, you know, launched in the same manner as a workforce transformation. So it's been nice to see that materialize. And of course we can't forget about safeguards, risk and trust. And so we're rolling out various programs down at a prompt level, to an organizational process level, to a procurement level to ensure that we have the proper business hygiene to do this in a more accelerated way. So I would say that the work continues and it's moving in a very fast pace. We're beyond early operational piloting stages and we're moving towards that first phase of saying that we are scaling AI. It's very difficult because of our operating model that I mentioned earlier, being running this like a portfolio company, if you will. It does have its challenges also trying to be the identity of state IT and centralized services as well at the same time. But that mindset that we have for, for speed and scale is really going to allow us to drive longer term adoption and sustainability with our investments here.
- [09:21]
A
Right. I'd like to get back to the challenges in a minute, but I think one mark potentially of a state's maturity level with AI is how it reacts or how its officials react when they hear the phrase agentic AI. Given that that's, you know, something of a buzzword, but maybe could indicate a more forward leaning approach to the technology. Where, where is Oklahoma at right now with agentic AI?
- [09:46]
B
The best way to think about agentic capabilities is what happens when the human has mastered being in the loop. Essentially, agentic capabilities requires humans to be above a loop and trust that loop enough. And so if you kind of roll that metaphor forward, what this means is that in order for us to have true scale and true value from capturing ROI through agentic capabilities, we need to start to think around automation and business process optimization first as a step one. So we're very focused in launching automation practices and capabilities and tooling that would help agencies rethink do they have the right processes and procedures today? Because the solve for agentic capability doesn't start with can AI do it? It can't. But it's just a matter of where does it do it and when do we kick the human out of the loop and let the human observe that loop. And so step one for us was making sure that there's proper business optimization. But part of that is we know those things take time to change. We can't wait for a good process to fully curate before deploying AI. And so the other piece that we're doing here to accelerate that adoption journey is to invite agencies into A lighthouse program wherein they can use pieces of a process, pieces of a journey, and really experiment with AI in a very safe state, Control tenant that we're calling beacon, so that they can actually see the potential. And what happens when you reimagine all the processes upstream that feeds into that and downstream that comes from that. And then finally, when we're thinking about this idea of agents, it's creating some very nice implications to the workforce. When agents work at scale, you essentially have new digital workers to manage. I think this is one of those moments where, you know, as a culture, we're probably seeing the last generation of leaders managing all human teams moving forward. What does that mean? How does that, how does delegation of authority work when agents are in the mix? And so we're actually going through that thinking right now to see what can be done by an agent and where is it appropriate to remove the human from the loop, knowing that at the end of the day it's human overseeing all the loops in which it used to be in. And so I think it's a nice change in tone, a nice change in our risk posture. You know, we have to stick the landing on who in the loop before we can even have these conversations. And we're being very diligent to make sure that we continue to provide that level of safeguards even as we thinking about humans managing a bunch of loops all at the same time.
- [12:15]
A
Yeah, and I think this is a good segue to talk about some of the challenges that you mentioned. That word challenges can mean a variety of things. One thing it might possibly mean is how people who work for state government are reacting to AI, whether they're receptive to it. I know there's, you know, we don't need to talk about potential workforce concerns, but that's something that comes up sometimes. So what are the sort of challenges that you concern yourself with when it comes to the state's AI work?
- [12:43]
B
We serve quite a few Personas throughout any given month. And the challenges really is predicated by the pain points that these Personas experience on a day to day basis. For our agency leaders, it's really about are they getting the right ROI value from the investments that they're making in these high dollars, multi million, potentially multi year projects tied to not only AI, but technology that now has AI as a component. And so part of that is us deploying a very clearer and more connected front door so that agencies know where to go for proper review, what vendor toolings to use and how can they experiment as they start to make these multimillion dollar decisions. But what's different about this door is that it not only has to connect technology data, but also legal, procurement, cyber and risk. We want to make sure governance arrives early on so that it doesn't slow down innovation downstream. And it's really important that we move with the right velocity, but not so f that we create mistrust over time. And so that front door has been heavily reimagined and we're seeing early benefits from that, not only through identifying duplicative contracts that otherwise would have happened without this, to ensuring that, you know, we're, we're positioning proper reusable capabilities early on so that agencies don't have to pay for things that the state already has. The second thing here is I'm going to go back to the workforce conversation because I think that challenge will always continue, especially for managers and supervisors who have staff who are afraid to use AI. Still to this day, we are seeing a significant amount of uptick in activities and prompt and really imagination and collaboration across the board when it comes to even desktop level productivity tools. But there's still this elephant in the room of will AI take my job? I think about a year ago that was a lot more pronounced. A year later. I think what we're saying is how can I use AI to protect my job? It's a little bit shifting from a more afraid to try. Now that they try, they want to keep playing to see how it can play in their advantage over the long term. And the way that we support that is through different models of collaboration, experimentation. Our CIO just recently launched 10 pods across the organization asking employees to take a step back and just reimagine the problem space that they have in front of them, build a fusion team and give them the right tools to in the right sandbox and say, how can we solve this better? And so I think there's been a nice encouragement of that at an employee level to really make sure that we are nurturing psychological safety where it counts. Because the idea of AI is still very much scary at the end of the day. And I think we've done a nice job curbing that fear and that we're feeding the appetite so that they can reimagine the problem space that they're dealing with on an everyday basis.
- [15:34]
A
Yeah, I think, I mean, not that I was ever terribly worried about AI in terms of it taking my job, maybe I should be, but seeing its limitations, having enough exposure to it and seeing that right now, as powerful as it is, there are Things it's good at, things it's not very good at. And efficiency becomes a matter of learning how to use a tool, just as with any other tool. So that's kind of maybe one. We're kind of brushing against the idea of misconceptions there. Are there any common misconceptions that you need to disabuse people of either within your organization or vendors you work with that, that, that come up frequently.
- [16:16]
B
I think the same ones that was here when ChatGPT was started to become unpopularized is still here. But I think the, the concerns around that and how we address it has gotten a little bit better and we're now just definitely dealing with the same themes of misconception, but just in a more clearer way. I think this, the first one that I still see is, you know what, what a lot of the perception around this is that AI is still magic. In other words, just buy the platform, buy the tool and wait until the POC is done and all of a sudden you have ROI and value. That's really not the case simply because in order to do that effectively you have to address a lot of non technological dimensions around the organization, process, procurement and legal. It's not a replacement for solving some of those things. And so I think that we've gotten better as an organization, the collective we of leaders across the nation who's driving some of these conversations, who are now part of my peer network. And I think the realization is that the accountability factors is still very much human, as is the ambition and it's really AI as the leverage. So I think that's, that's the evolution of, I think that that misconception around AI being magic, it's not just about make your process better, it's about ensuring that your ambition is still right size for what the technology can do now without forsaking things around process and data, legal and procurement. I think that's, that's one of them second piece is I think that we are, we've entered this, this period where, you know, when you think about AI, especially in a government context, the stigma of governance is still very, was, was very loud a few years ago. And the idea that governance slows down innovation, I would say no longer holds. I actually believe the opposite. Governance is what allows this to scale, which is a different placement for a word that traditionally never belonged in a conversation about how to go fast without governance. You might be moving quickly at first, but eventually you'll hit risk, duplication, procurement issues. And so adaptive governance is really the key to AI success. And I think that's one of the many things that has changed as we think about the future moving forward. And for us at the state, we've created Accountable innovation operating model aiom what that for us is a just in time, just enough governance layer from conception to scale that encompasses all the areas that needs to be covered to minimize our risk without slowing us down. And without then I think that finally, when we think about what's really important in moving forward is that the people aspect to this will always be there. People will always be concerned. Better use of AI is often about helping people do their jobs and better. But I think that we now have to grapple with the fact that can we trust, we have to trust it beyond more than just basic level productivity. And that's really the next frontier. And I think the trust level would require that different jobs be created and different flattenings of roles be created as well. And that is, I think something that most people don't realize is that AI doesn't necessarily reduce roles. It displaces tasks, flattens the organization, creates new job families. And I think that workforce transformation is, is still in some ways elusive for most organization. For us, I think we're dealing with in a very pragmatic way of ensuring that we're able to streamline journeys and processes before looking at the practical application of what job families are better suited for that new future state. And so I think we're still figuring that out, but we definitely have a point of view on making sure that that's done in the right way, in the responsible way.
- [20:10]
A
Yeah. Now, in terms of, you know, you could answer this as broadly or as specifically as you like. You know, you talk about job families or, or are there specific use cases that you think are the sort of best way or the most promising way to apply AI in state government, what do you think those might be?
- [20:31]
B
I think for, you know, if you ask all the vendors out there, they're, they're going to hit you with a lot of things around agentic capabilities to drive customer service, how to predict fraud in a much better way, how do we build our digital workforce? And I think that's all still true. AI still have those promising use case that the last two or three waves of technical innovation held for us. What I'm finding is that boring things are still sexy. So when you give employees the ability to reduce paperwork, helping people find answers faster or giving time back, it allows them to work on those promising use cases. And so for us, I think the portfolio mix of priorities really is evenly balanced across the board work. You know, on the one hand, we have a chunk of our portfolio that's really focused on return to employee for that very reason. We want to make them more productive. We want to make them feel more empowered, want to give autonomy where it makes sense to really do their job faster. But on the flip side, those promising use cases is against a measurement of what the industry analysts will call return to future. But for us, it's more important around, well, how do we return things back to citizens and how we return things back to the state? And so the use cases for us really is categorized within those four frame definitions of value. What can we do with AI to serve citizens better? What can we use AI for? To make legislative processes run faster? And I think if we get all those three focusing on our employees running government faster and doing better services, we have a very strong predictor of what return to Future is.
- [22:14]
A
That was Tai Phan, Oklahoma's chief AI and technology officer. That's it for this episode. The Priorities Podcast is a production of Scoop News Group in Washington, D.C. adam Butler and Carlin Fisher put it together. I'm Colin Wood. Thanks for listening.