---
id: "workshop:stakeholder-engagement-in-ai-implementation"
type: "InnovateUS workshop"
title: "Stakeholder Engagement in AI Implementation"
url: "https://innovate-us.org/stakeholder-engagement-in-ai-implementation"
date: "2025-10-06"
duration: "60 minutes"
language: "English"
series: "AI for Law Enforcement"
instructors: ["Michael Navin"]
recording: "https://www.youtube.com/watch?v=5gz1_TZooyY"
transcript: "automatic, from the recording"
source: "InnovateUS (innovate-us.org)"
licence: "CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/), as stated on innovate-us.org"
---

# Stakeholder Engagement in AI Implementation

InnovateUS workshop · 6 October 2025 · 60 minutes

Successful AI implementation in law enforcement requires active involvement from a broad set of stakeholders, including internal departments, oversight bodies, and community representatives. This session will help executives develop stakeholder engagement strategies and processes that support trust, transparency, and shared ownership throughout the AI lifecycle—from problem definition to deployment.

## Learning goals

- Develop stakeholder engagement strategies and processes that support trust, transparency, and shared ownership throughout the AI lifecycle
- Understand the stakeholder engagement process from problem definition to deployment

## Speakers

- **Michael Navin**, Principal Court Management Consultant, National Center for State Courts

Part of the series [AI for Law Enforcement](https://innovate-us.org/workshop-series/law-enforcement).

Recording: https://www.youtube.com/watch?v=5gz1_TZooyY

## Transcript

Automatic transcript of the recording: names and terms can be misheard, and the times can be a few seconds off. Each time links to that moment of the video. The transcript alone: https://innovate-us.org/stakeholder-engagement-in-ai-implementation.txt (plain text) and https://innovate-us.org/stakeholder-engagement-in-ai-implementation.json (every segment with its start and end second).

**Michael Navin**

[0:02](https://youtu.be/5gz1_TZooyY?t=2) Great. Let me just go through the steps here and share my screen. All right, great. I think everybody can see this. Well, thank you, Jess. Thank you for the introduction. I'm happy to be here to talk about stakeholder engagement in AI implementation and why it's so important. I mean, it's important in any technological implementation, but there's some differences in AI implementation that are important to go over. But a number of the things that I'm going to talk about, they're relating to courts. But in the public safety sector, a lot of these can definitely be transferred. What I'm telling you can be transferred to whatever area you are in. Although I'm talking strictly about courts, there's nothing here that you can't use.

[1:06](https://youtu.be/5gz1_TZooyY?t=66) Just to give you a brief overview of the National Center for State Courts. The National Center for State Courts has been around for over 50 years, and our mission is very, very simple, to make courts more fair, effective, and efficient. And we're also a resource hub for supporting state courts. We provide research, consulting. We help courts with their data and tech solutions for courts as well. And really, we partner with judicial leaders, all of the chief justices of the state supreme courts, as well as state court administrators. And we also offer a lot of court management training, which is a very highly regarded certification program that we offer. And we also support a lot of global justice reforms. We extend court improvement work internationally. So we really have quite a diversity of clients that we work with within the courts.

[2:12](https://youtu.be/5gz1_TZooyY?t=132) As Justin said, just briefly, my name is Michael Dabner, Principal Court Management Consultant with NCSC, and I've been here about two years, but worked about two and a half decades in various technological situations between two different courts in the Detroit, Michigan area, implementing e-filing systems, case management systems, and the like. So just to give you a brief overview of what we're going to be talking about today, where we're headed, we'll start with why stakeholder engagement is so critical for adopting AI in operations. And that's the important piece too, is that this is about operations within the courts, not necessarily what courts may see coming in front of a judge. And so we're looking at kind of the behind the scenes. And then I'll walk you through the AI lifecycle and where stakeholder engagement is important in defining the problem, as well as the requirements for implementation. And then we'll look at who to engage, and then some tools and some case studies and some examples that are real that have happened in the court to show exactly how they work out in the more pragmatic real world.

[3:27](https://youtu.be/5gz1_TZooyY?t=207) So why stakeholder engagement? Why does it matter? When it comes to the courts, courts have a unique mission. We need to keep our legitimacy. We have to keep up with fairness, and we have to keep our trust with the public. And courts are using a term, when it comes to AI, they're a high risk entity, meaning they really can't afford AI failures, because if something doesn't work, it makes the news. And this is the case, I think, with any other public safety entity out there, whether it's police, 911 operators, corrections. There's not a lot of room for failure there, especially when it comes to AI and dealing with a lot of the issues with AI that can come up. If the staff or public feel excluded, then trust erodes.

[4:29](https://youtu.be/5gz1_TZooyY?t=269) And so engagement ensures that AI meets the real needs. And in AI and in operations, it affects clerk and staff. It affects the processes, how things move through the actual processes step by step. And judges and administrators, the oversight, making sure that those processes are being followed. Where are the pain points? Where are the bottlenecks? And then the users, the public users of the courts, the self-represented litigants and And looking at those. attorneys having access to the courts. Those are vital stakeholders as well. And so without engagement, without engaging all of those type of stakeholders, you're tutoring on having failed adoption, mistrust, and wasted resources. And again, courts, they don't have just bloated budgets. And so they need to be very careful with how they spend their money.

[5:31](https://youtu.be/5gz1_TZooyY?t=331) And again, that's the same way with other public safety entities as well. Everybody within government has specific tightness on resources. And that's not just monetary, but that's also staffing as well. So with that, I want to start with a poll. And I want to see just the first question is, I just want to get an idea of who you all are. What entity are you from? And if you can just take a minute to answer that question, please. Okay. A lot of law enforcement, a lot of other. I guess I should have came up with some additional choices to potentially reduce the other, but that's okay. All right.

[6:31](https://youtu.be/5gz1_TZooyY?t=391) Thank you for that. I appreciate understanding where you all are from. So let's go through some of the basically where stakeholders engage in the life cycle of court operations. But again, if you look at this life cycle on any technology project, whether it's AI or not, it follows probably the same process. And the very first spot is the very first step is really defining your problem. What problem are you trying to solve or process are you trying to enhance? It's so important to identify that or those, coming up with where are those pain points? Where are those bottlenecks? Where are the frustrations of staff as well as the public that come and see it?

[7:35](https://youtu.be/5gz1_TZooyY?t=455) And it's so important to get the stakeholders involved at that point, all of your stakeholders, the frontline workers, the administrators, the executives, human resources, procurement, really identifying that and having some humility to say, okay, here's where the problems are. It's very, very important to get that. And because the thing is that defining your problem right now, it's going to make it so you're looking for where are the problems and not where can AI, where can we use AI? AI is simply a tool in your toolbox. So it may help something that you have a problem with. And that's fine. And that's okay too. Or it may not. But it's very important to start with where your problem is. And then that leads to the data readiness. You know, AI is only as good as your data. If you have bad data, you're going to have bad AI. So, you know, looking at what is the status of our data on what it is that, you know, where those problems are.

[8:41](https://youtu.be/5gz1_TZooyY?t=521) Are there missing fields? Are there missing, is there missing data on individuals or entities? So that's very important. And then building those, you know, continuing on with those requirements and getting the stakeholders involved in building requirements or if you're looking at an off-the-shelf product, you know, looking at to see what is the functionality of something, getting every stakeholder involved to take a look at that to make sure nothing's missed. Then doing a pilot once you're kind of developing. And, again, I know this is very high level. But when you're getting to the point of doing a pilot, having, you know, your stakeholders involved in implementing or launching that pilot, as well as the evaluation, the next spot, you know, having those people working on that pilot, maybe a small section of staff that are simply working on those that are maybe a little more of your subject matter experts, and then getting back that evaluation, having those stakeholders provide you with information back in order to see, okay, what's working, what isn't working, where there's some bugs, where there's some things that are not quite clear.

[9:52](https://youtu.be/5gz1_TZooyY?t=592) And then from there, developing, you know, making some changes on that and then deploying it more, you know, to the entire organization. And then, you know, from that, you continue to, you know, get information, monitoring it, getting feedback from those stakeholders to see, okay, what else is out there that we might be missing? And where else can we take this, that feedback of we've deployed this, now where can we scale it to something else? Where can we build off of this? And getting some feedback from your stakeholders as to where you need to go on that aspect. So to give kind of an example, you know, who to engage, and again, this is in court operations, but, you know, the problem, you know, how do you do that? You know, well, you get your stakeholders together and you have listening sessions and you kind of define, you know, define success. And the listening sessions can just be people talking about the things that give them a lot of stress, their pain points and things like that. It's very important to do this.

[10:53](https://youtu.be/5gz1_TZooyY?t=653) And I'm going to give some, again, near the end, some real examples of how that has happened within the courts. The data, you know, like I said, you know, data ownership workshops and looking at the privacy review. And that's an important piece too. When you're looking at AI, what data are you using? Is there personal identifying information that's there that you want to look at to make sure that that's protected? Requirements, designing those user stories, you know, really pie in the sky, what do you want it to look like? And then having some acceptance criteria. Then procurement, this needs to be looked at obviously right from the get-go, you know, because it's important to look at other vendors, especially if you're thinking about using an outside solution provider to build something for you or that's off the shelf. You need to get procurement involved in that right away. Then piloting, the training, making sure that people understand how the application works, not only how it works, but also how it fits into the workflow of their job.

[11:57](https://youtu.be/5gz1_TZooyY?t=717) A lot of times, you know, applications are just put onto staff and they're like, here you go, here's the training on it. But they're not really taught how to actually use the application within their workflow. So it's important to include that and get that feedback from your stakeholders as well. The evaluation metrics and bias checks, really important with AI because again, if your data is biased, if your algorithm is biased, your AI is going to be biased. So you need to have your stakeholders evaluating that as well. Then deployment, having a communications plan, getting your public information officer folks involved as well. And finally, monitoring, you know, that's everybody, you know, getting feedback from anybody from the outside, from the public, to your judges and court administrators, to your support staff. So we're going to now take a second question. What would be your one biggest pain point from the list below? Take a few minutes to select.

[13:28](https://youtu.be/5gz1_TZooyY?t=808) Yeah, that's, you know, and that's pretty consistent. I think, again, across all entities and I've been in other conferences with other public safety entities. This is almost exactly the same responses that we're getting, especially that lack of resources, you know, whether it's, you know, budget resources or whether it's staff resources. A lot of people, you know, since COVID, there's a lot of turnover, especially at entry And that can be very frustrating when it comes to really trying to, you know, train level jobs. folks, trying to upskill folks. It takes a lot of resources to fill in, you know, to train those resources coming in. And then, of course, the inaccessible data and so on. So, you know, what this really shows is that, again, you know, really kind of the first conversation about AI is not to talk about AI. It's to talk about these type of things, your lack of resources and getting those stakeholders together to do all this. How do, you know, the issues with upskilling staff, inaccessible data, inconsistent information given the public can be so frustrating when four or five people are giving four or five different answers to the public.

[14:35](https://youtu.be/5gz1_TZooyY?t=875) And there are solutions out there, you know, for AI that you can build from that can help with that. But, again, you need to have your stakeholders involved to get through that. So where do you start? And so this is where step one is. And this is defining kind of your starting point of your collective stakeholders. And this is just the really the starting point. This is the start to get things started. So within a court, you maybe just want to have four people, these four people, and maybe it's more than that, but just to actually have a conversation as to how do you get things started? How do you actually start digging a little bit into the dirt or maybe just determining where to build? So within the courts, that would be a judge, a court administrator, the chief information officer, and your general counsel. All areas, you know, the judge to kind of be a champion and the court administrator to know about the processes and the chief information officer there to be able to, you know, know the technology, and then the general counsel there to be able to look at the rules and the regulations and laws that could become issues right out of the gate. So very important to just start with this.

[15:52](https://youtu.be/5gz1_TZooyY?t=952) And that leads to, you know, defining really the scope of the AI, you know, very high level, you know, like an example, like we want to develop a roadmap, you know, we want to have a strategy and a roadmap as to where AI could be implemented within our organization. So that's, you know, an overarching high level aspect, you know, of developing that AI scope. And so, you know, AI can help, you know, and this is the start of that. And, you know, you'll dive deeper into more of the details of the scope. But right now, defining that kind of mission on where AI can be used is really a critical step, second step in the whole aspect of AI implementation. Step three, so you now are able to branch out and okay, you've got kind of this idea of where your scope is with AI from the collective stakeholders. And now you want to create a committee, a governance committee of other stakeholders. And that could include, again, this is not all inclusive, executives, your legal unit, again, your general counsel, your CIO, your CISO, someone from budget and finance, human resources, very important since things or AI is going to change how jobs descriptions are and how people work, procurement, your public office, your public information office, and then your frontline supervisors. And then if you have a strategic planning division, you know, all of these would, again, not totally inclusive, but these are kind of some of your main players that would be part of the governance committee. And the governance is so important because that's setting the rules in how AI is going to be governed, you know, what happens when they're looking at the risks and looking at where AI could be implemented. This is where all that's going to be there. So these are kind of, this is your main governance committee that

[17:55](https://youtu.be/5gz1_TZooyY?t=1075) you can have. And then, you know, within that, NCSC has established this kind of governance framework and we've developed this out into five different committees from that governance committee. And the one committee is vision and leadership. And this is kind of that team that's going to, you know, set guiding principles and establish policy. And they're kind of the maestro in the orchestra. And they're just going to provide the overall guidance for the other committees. And when the other committees have questions that they need guidance, this is the one that they would go to. Stakeholder engagement and communication, you know, they're going to engage in their, you know, internal and external stakeholders. They're going to set expectations and they're going to ensure that, you know, everybody is involved and everybody's aware of what is happening in AI, both internally and externally. Education and skill development, you know, building that AI literacy. And that can be very fast moving because I think every day I come across a new AI term that I've got to look up to see what it means. So training the stakeholders, like I said, not only training potential users on AI, but also on how it fits within their workflow.

[19:07](https://youtu.be/5gz1_TZooyY?t=1147) Very, very important. And then technology architecture. These are really your IT folks, you know, how is AI going to integrate with existing systems and really addressing security and privacy. By adding AI in, that's adding another level, another quote unquote vulnerability for hackers and for adversarial attacks and things like that. So you have to keep, there needs to be, that group would look at those type of issues. And then court operations. How is AI going to impact the processes and where are those pain points and bottlenecks within the processes? And that's where you're going to get that information from. So you have those, you know, five committees, and then those five committees will have a representative on the main governance committee that will report back. And, you know, they each have their different, you know, different but similar goals and outcomes, but then their representatives go back to that main governance committee to really get an idea of what's next and make kind of final decisions. And, you know, this is a good stakeholder engagement example of some stakeholder groups that are out there. It helps us kind of tailor how we engage different stakeholder groups throughout the AI lifecycle. So you have frontline staff and court administration are both high, they have a lot of influence, and they're also highly impacted. And they should be deeply involved in the requirements definition, the pilot design, and in the governance.

[20:45](https://youtu.be/5gz1_TZooyY?t=1245) IT is critical for, you know, data readiness and integration, but they shouldn't drive use cases alone. And then you have procurement and communications are training and training are often a lot of times are brought in too late, but they need to be at the table right at the first meeting, you know, early engagement avoids those delays and can help with some resistance. They have judges in their administrative role. Often, they provide oversight and policy alignment, but for operational AI projects, they typically aren't the primary end users, but they still do have, you know, they still want to be informed as to where things are going. The key takeaway in this whole aspect is that engagement isn't a one size fits all, it's strategic and role based. And that's the important thing to keep in mind when it comes to stakeholder engagement.

[21:47](https://youtu.be/5gz1_TZooyY?t=1307) So in the court operations context, courts are facing rising caseloads in a lot of different case types. You know, COVID is now over, we saw a number of years, courts saw a number of years of decreased cases because of caseload, because of COVID, but those have all, for the most part, come back. And some of them are even rising higher than they were prior to COVID. And they're also facing staff shortages. Again, with COVID, the great resignation, it's causing a lot of turnover, again, as I mentioned earlier, with especially at those front line entry, you know, those entry level positions. And filings are becoming more complex, and maybe even getting more complex because of AI use. That's a completely different webinar. But you know, the use of AI, especially among self-represented litigants, you have issues there as well as to whether the cases are cited correctly, and there's no hallucinations. So there's complexity there that courts need to watch out for. In higher public expectations, people are using AI every day in their everyday life. There's no doubt about it. I would bet that every single one of you, in some ways, is using AI on a daily basis. So when it comes to AI and operations in the courts, AI could help automate filing triage. Where do cases go when they're filed? Because that is such a pain point, very time-consuming. Streamlining, scheduling, you know, maybe scheduling things so that there's more of, you know, on a daily basis, there aren't kind of swings, you know, where you have 20 cases one day and three days the next, and AI could help that. And reducing repetitive tasks, and that's what AI is really, really good at doing, taking out kind of those repetitive tasks so that staff have more time to work on the more complex issues that are out there. So, you know, bottom line is,

[23:52](https://youtu.be/5gz1_TZooyY?t=1432) again, adoption, all of this, though, adoption is going to fail without that stakeholder engagement from the very beginning and at each point in the life, in the project life cycle. Step six, focus groups and surveys. And we had a great use of this when we worked with the District of Columbia Courts. Again, I'll talk more about that in a little bit. But the focus groups and surveys, you know, surveying your staff, surveying your executives, surveying every aspect of your organization to really ask, and it can be anonymous, it's just, but at the same time, maybe it's not, you know, but if you have a very good culture of allowing people to be open and talk about some of the things that are problems within the organization, this can be incredibly helpful. And, you know, even if you don't have that, then I would suggest starting with that, with really opening it up to where we really, you really want to find out where there are these pain points within your operations and making people feel comfortable saying those things. And so you can find the pain points for employees, the pain points for service recipients, those that come to use your services. Those, you know, what are labor intensive and repetitive tasks that just kind of make people just not all that, kind of make people feel as though they're just kind of turning a widget or making a widget every day, you know, trying to look at where, what are those and how can we reallocate staff to not do those repetitive tasks. Error prone. Errors, you know, if you make that error right from the very beginning, it only takes a second to make an error, but down the line it can cause problems and it can cause hours of work to correct an error. And so it's good to find where are those at and you can use metrics for that too. You know, hopefully you have management systems that you're able to kind of see where those errors are at and then how can you, and then AI has the ability to

[25:54](https://youtu.be/5gz1_TZooyY?t=1554) really help reduce those errors in order to make things move more smoothly for all situations instead of just most. So now another poll question. If your organization were considering implementing a new AI tool to support operations, which stakeholders should be in the room when defining requirements? Select all that apply. Yeah, pretty much. It was kind of a trick question because they should all apply. Everything, every single one of those should be, should be at the table when defining requirements.

[26:54](https://youtu.be/5gz1_TZooyY?t=1614) So yeah, and it's, it's pretty much what you all are saying as well. It's just, it's vital. You can't, you can't leave anybody out. Okay. So step seven. So you have, you did your surveys, you did your focus groups, you did maybe some additional interviews with people to find out where those pain points are. And each of those separate subcommittees within the governance, they come back and they, they talk about, you know, they brainstorm on, you know, to see where, where are there some systemic issues across all of those, at all of those committees and recurring pain points. And, and then they start thinking about, okay, how could AI solve or enhance and maybe doing some research on, on other entities that have used things to, to help along with that. Metrics, what metrics could be used to, to really be able to show something, something data-wise to be able to show, okay, this is where we're at right now.

[27:57](https://youtu.be/5gz1_TZooyY?t=1677) We have a, you know, on this particular filing or this particular issue, you know, it's, it's prone to have an error rate of 70%, you know, okay. You have that before metric. And then, you know, the goal, the hope is, is that with an AI solution down the road, you can then be able to compare, did that AI solution resolve or help the error rate? Did it help it reduce the error rate or the, or the, or the amount of time? So that's really important to have some kind of metric and it doesn't have to be error, but it can be, you know, is there a cost savings or is there a, just amount of time being spent on something that has been reduced? And then you take those recommendations to the full committee. And when all of them do that, they should be able to just kind of start vetting to say, okay, where do we go from here? How do we find kind of a first AI project? Something that is low risk, has a high value, a high return, and, and also, you know, some other issues like, you know, something that's an easy win. And that's, what's so important about AI, you know, especially on your, your first AI project, what's an easy win. So you can keep the momentum going, something you can take that and scale up to, to, to, you know, to whether it's, it's something separate, maybe it's a chat bot, like, oh, we're just using a chat bot on this issue, but it works so well.

[29:19](https://youtu.be/5gz1_TZooyY?t=1759) So now we can, we can scale it up to, to another part of operations. So all of those things is, is what you would, you would hash out once you have the brainstorming and the recommendations back to that full committee. I'm going to give you an example of email, e-filing automation, and this, this has to do with, this was, this occurred in Palm Beach County, Florida, and it also, it's the same AI solution provider that is working also in Tarrant County, Texas, as well as Dallas County just recently. The challenge was, so this is an e-filing system and how the e-filing system works is a filing comes in and someone has to actually Should we go in and accept the filing or forward it on to wherever it needs to go? But there's a lot of filings that are very kind of repetitive and don't really need a lot of review. And clerks were becoming overwhelmed with a backlog of complex filings. But they had to keep up with kind of the more everyday, repetitive filings that they needed to review and move on.

[30:27](https://youtu.be/5gz1_TZooyY?t=1827) So the solution was, hey, we can automate routing of some of these filings and then reallocate those resources to work on the complex filings. That was the goal. So they got the stakeholders involved. Clerks looking at the usability of the solution. IT looking at the integration of this AI solution along with their e-filing and case management systems and their document management systems. And then leadership, making sure that policy aligns with all of those aspects. And of course, they had metrics, data, looking at accuracy, backlog levels and time of acceptance. And the result was eventually they worked their way up to having 150 documents that were automatically, when they came into the system, they were automatically routed to where they needed to go and with no human intervention. They end up having about nine bots, which equals about 45 full time employees. And right now, 42 percent of filings are never reviewed by a human.

[31:29](https://youtu.be/5gz1_TZooyY?t=1889) And so they were able to take. Now, I want to be clear. Nobody actually lost their job. They simply took those folks and reallocated them to somewhere else. In the public entity sector within the courts, courts are there's always work to do. And I think that's the case in any other public public safety entity. There's always work to do and they're just going to move people around and reskill them. So it's kind of a good place to be in to automate these repetitive tasks. And it works successfully and they're continuing on in scale and continuing to scale this to other case types as well. So next poll question, if you had to choose just one, where would you like AI to be leveraged in your organization? And I'm going to pick this question really quickly again.

[32:45](https://youtu.be/5gz1_TZooyY?t=1965) Very, very doable. Invoice processing, there's a lot of bots out there that can help you through. By keeping a human in the loop, processing various aspects of invoices to allow people to do other work. Summarizing and drafting documents, that's a big one too. And I think that's becoming more and more popular. So there's not a lot of what, out of what you selected, there's a lot of low risk, I think, involved in some of these, to a certain extent. So that's great. So I wanted to kind of give a spotlight on, IGES is a organization that I work with closely, Integrated Justice System. And they have released earlier this year a playbook. You can access this at the website again, right here. Again, you'll be getting a copy of this PowerPoint so you'll be able to click on this link when you get it. But it's a playbook of about 13 different plays.

[33:48](https://youtu.be/5gz1_TZooyY?t=2028) And what a play is, is that it's a segment of AI that you can focus on, that allows you to focus simply on that one, quote unquote, play. So there's a play for governance. There's a play for stakeholder engagement. And the stakeholder engagement is play six. And it talks about engaging stakeholders earlier. Building that feedback loop for implementation and assigning those governance roles. And then documenting decisions. And that's a big part of the stakeholder engagement. But it's a bigger play, in a bigger playbook, like I said, of about 13 different plays. And they ask things. There's things like governance, workforce development, as well as use cases in different public entities that are out there. So it's a great resource. I highly suggest that you go there. And it gives you really basic one or two pages of resources that you can use, checklists, and then key questions that you should ask at each point.

[34:49](https://youtu.be/5gz1_TZooyY?t=2089) So a great resource to look at that is right at your fingertips. The NCSC, we have just recently released an AI readiness tool. And it's a foundational tool that helps courts build the three building blocks right now. And stakeholder engagement is vital in each one. And our first step is that building the foundations, getting your stakeholders involved, developing an internal AI use policy, your guiding principles, and those type of things. And start really evaluating the processes. And then moving on to the second building block, which is developing that and implementing a first AI project. And then we go on to provide additional information on building off of that, providing that feedback loop. And we also provide a lot of use cases, at least with courts, showing how AI was done as a first AI project and how all of the stakeholders are involved in it.

[35:54](https://youtu.be/5gz1_TZooyY?t=2154) So it's a great resource. I'm going to show you near the end a QR code where you can access that. But again, stakeholders are vital to that readiness, to really pull that AI readiness together to move forward with your AI journey. Some other examples that NCSC has worked on, I mentioned the District of Columbia Courts. We worked on a year-long project developing an AI strategy and roadmap. This was a great project, building that initial collective stakeholder group. And then from there, building out smaller groups, dealing with a number of different committees that were very similar to what I listed out, to vision and leadership. We assisted with that committee development, AI use policy development, focus groups, and then eventually coming up with a first AI project recommendation. All of this revolving around getting the stakeholders involved that needed to be involved. And they were such a great court to work with, so open to understanding their pain points and their bottlenecks, and really looking to see where AI can solve some of those.

[36:59](https://youtu.be/5gz1_TZooyY?t=2219) The Supreme Court of Georgia, we also worked with them on seeing how is AI being used around the state. And we came up with some recommendations and some AI guidelines and safeguards in a report that we just released just a couple of months ago. Some ones that we're currently working on right now, the timing could not be good enough on this Idaho judicial branch. We're working with them on developing their AI governance and data governance, because data is so important. These are great projects to work on together. And we're at the point where we created that big governance committee, and now we're forming the smaller committees to get all the stakeholders involved. And it's great we have your normal HR folks, but we also have academia involved, and we even have some solution providers as part of that as well, too. And it's based on that AI readiness project that I talked about. We're walking through those steps. And we're doing some engagement, too.

[38:01](https://youtu.be/5gz1_TZooyY?t=2281) You can't leave out solution providers. And solution providers, and we're engaging with an appeals court in Texas, helping them engage with solution providers to help develop a potential solution with case categorization, a.k.a. case triage. And we're working with them on the RFI development, and then that solution provider outreach, getting that engagement from them. And it's going really, really well. So with that, NCSC has these type of expertise. We're now developing the AI governance processes for them and the procurement. And then in solution implementation, we're actually working with the state of Alaska, helping them develop a chatbot in probate cases. So it's a very exciting time to be working on all these. And you have a starter kit from what I've given you in those first seven steps.

[39:02](https://youtu.be/5gz1_TZooyY?t=2342) And I think that's going to be a good start for you to do it. And we're always around to help you. And on some of those other, sorry, I've got to move some things around. These are the codes that I was talking about. We have the IGES AI Playbook on your right, great resource. We also have a policy consortium that we're working with, with the Thomas Reuter Institute. This is all great, free information for you. Again, I know it's focused on the courts, but every aspect could be incorporated with other public safety entities. We have a monthly webinar that we provide information on use cases and governance and just those kind of ethical issues with AI. It changes every month. So please feel free to go there and sign up for those webinars because they're really, really, really helpful and really useful. And then the AI readiness that I was talking about earlier, the three steps, the foundational aspects of getting you to that first AI project and hopefully being able to scale it even more than that.

[40:11](https://youtu.be/5gz1_TZooyY?t=2411) So three very, very good resources that are at your fingertips. So just quickly, you've seen how AI governance and stakeholder alignment determines success. So most jurisdictions and most entities, they solid awareness. They'll watch something like, wow, that's great. They understand the need, but they don't know how to move into that structured planning. Planning is where stakeholder engagement really becomes operational. You dip your foot in that pool and setting priorities and roles that determine the downstream success. And then that leads to the execution. Execution succeeds only if early engagement is strong. Weak foundations lead to fragmentation and stalled initiatives. So strategic implementation builds on early engagement. I can't stress that enough. What are some common next steps after you have that governance alignment of all your stakeholders? After governance structures are in place, jurisdictions and entities typically move into workshops to clarify roles and map structures And the next layer is an AI readiness assessment, and build consensus.

[41:14](https://youtu.be/5gz1_TZooyY?t=2474) evaluating baselines, identifying policy gaps, and producing structured recommendations. And finally, many advance into the pilot planning and implementation, where the use cases that you've come up with are tested and governance checkpoints are applied and scaling strategies are refined. So these are some next steps you can take after that governance and stakeholder engagement aspect are fit in. Some other strategic next steps. Effective engagement starts with identifying the right stakeholders and making their roles explicit. That's really the importance about governance and identifying right stakeholders, making sure that everybody knows what their role is. Misalignment here can derail everything downstream. So clear communication and feedback loops turn governance from static committees into active decision-making structures. So embedding engagement into planning and implementation ensures it drives real progress rather than just becoming a one-time exercise.

[42:18](https://youtu.be/5gz1_TZooyY?t=2538) The set it and forget it ways of technology, they're over. You have to continue to develop and work off of what you've already implemented. It's a constant loop, and it needs to be constantly looked at as to what you've built and how it can be made better and what else you can solve with that AI solution that you've implemented. So last question. What is going to be your next step after you leave this webinar?

[43:22](https://youtu.be/5gz1_TZooyY?t=2602) And unsure, need more internal discussion. It's OK to be unsure. I was at a conference recently, and I was talking about AI with somebody in a court, and they said, we're behind. And I said, no, no, no, no, you're not behind. AI is moving quicker than how the human brain can comprehend right now. It's so evolving so fast. And I really need to stress this, that don't ever feel like you're behind, and it's OK to be unsure. You take the amount of time that you need to make sure that there's so much risk involved, especially in the public sector, that you take the time you need for everybody to be comfortable with where you're going and the direction you're going. Take it step by step. There's nothing wrong with that. Don't try to implement AI just to implement AI. Make sure you're solving a problem or enhancing a process. And I really can't stress that enough. Don't try to beat everyone else to the punch, because that can cause issues.

[44:26](https://youtu.be/5gz1_TZooyY?t=2666) And so being unsure, that's fine. And I'm hoping some of these resources That's great. that I just provided you can help guide you a little more on becoming a little more sure in what you're doing. Bottom line, effective stakeholder engagement builds the foundation for sustainable AI adoption. Please keep that in mind. And it's so valid. And I thank you for taking an hour of your time today to talk with me. I'm going to answer any questions. But if you have any questions, please feel free to reach out If you have any questions outside of this, please don't hesitate to reach out to me via email. to me. And I'll be happy to answer any questions that you may have. Thank you very much.

**Audience**

[45:06](https://youtu.be/5gz1_TZooyY?t=2706) Thank you. Thank you. Great. Well, Michael, thank you so much. I'm going to stop the recording. Thank you. Thank you. Thank you.

* * *

Original page: https://innovate-us.org/stakeholder-engagement-in-ai-implementation · Licence: CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/), as stated on innovate-us.org