AI in Courts: Where Are We Really, and What Should You Do Next?

Brendan Hughes: Hello, and welcome to Scaling Justice, a video podcast by equivant that operates at the forefront of the justice space. I’m Brendan Hughes, Director of Marketing at equivant. All of us at equivant are dedicated to delivering innovative solutions to simplify justice. Today’s episode of Scaling Justice is focused on the topic that is everywhere when people speak about technology in courts and justice agencies, and that’s AI. If you’re like me, your head might be spinning, seeing headlines of AI and core technology, seeing lots of promise reading about what’s possible and cool features, but also might be feeling like you haven’t really seen or heard about AI and core technology that’s actually being used. So today we’re joined by experts that are going to cut through the noise, provide an honest, grounded view of what is really happening in the justice sector with AI technology. 

And if you’re feeling behind, they’ll let us know if you really should be feeling behind. But if you are feeling that way, they’ll provide some guidance on practical next steps anyone can take today, regardless of size or budget. So let’s just jump right into it. Let’s welcome our guests, James and Leigh. Could you guys maybe introduce yourselves and provide a little background? James, why don’t you go first? 

James Young: Sure. Happy to be here. James Young, a partner at Guidehouse. I’m responsible for our state and local technology. I have just under 30 years of implementation experience all in public sector. Actually started as a Java developer and worked my way up through solution architecture into business development and managing large enterprise teams. So a lot of experience in state and local and justice and court going back longer than I’d like to acknowledge. 

Leigh Sheldon: And my name is Leigh Sheldon. I’m a partner at Guidehouse and lead our AI and data practice across state and local government. So I have the privilege of working with James every day in terms of how we show up and provide services to our respective clients. And, you know, it’s been a really rewarding past few years. I’ve spent the last 16 years serving clients across the full data and analytics life cycle. But in the past three years, you know, you read the news and something’s on the news about AI. It’s a really fun, rewarding time and helping navigate clients through how to make sense through the noise. 

Brendan Hughes: I appreciate you guys both being here. Lots of experience in this space. And let’s just jump right into it. When you talk to court leaders around the country generally, what’s that sentiment about AI right now? Enthusiasm? Anxiety? Something in between? Leigh, why don’t you start us off? 

Leigh Sheldon: I would offer with somewhere in between, I find that some are already and have been having the conversation for the past year, notably thinking through how do we get started, where should we invest more of our initial time. And then there are many others who, they’re not there yet. They’re actually focusing more around improving data quality and that’s their objective before they’re actually moving towards application of more advanced technology. So it really does cross the entire continuum. But at minimum, I think everyone is starting to think, what should we do? 

James Young: I think Leigh hit it right on the head. You know, the conversation has moved from should we be using AI to how do we use AI? And the first step of that is making sure that they’re looking at are they going to be able to manage the ethics, avoid bias, being able to explain the outcomes. Right. So a lot of people are still in those early stages focused on the how do we use it. And then to Leigh’s other point, the data, cleaning up the data and making sure you have access to the data so that your AI engines are well informed. 

Brendan Hughes: Great. And you know, that leads me to my next question, which is a lot of people are thinking about it, starting to move down the path of adopting it in your sense and your feeling right now, how many people are actually in the real adoption phase? And maybe, James, you could start on this one. 

James Young: Yeah, I mean, I’d say it kind of varies court to court, but overall, I’d say they’re still very early in the journey. Awareness is kind of where things start. Right. And I think that’s high. As we talked about with people thinking about how do they use it. I wouldn’t say there’s a lot of people pushing into production deployments because of the sensitivity around Quartz and then having to get it right. You don’t see the same mentality that you do in other areas like in commercial, where it’s fail fast. 

Brendan Hughes: Right. 

James Young: With courts, you can’t fail. Whatever you put into production has to be very thorough. So we’re seeing a lot of courts start with, you know, their governance policies, creating a task force. And if they are doing things in production, it’s usually starting with small pilots just to kind of get a sense for how things are and where they can be leveraged. They’re starting to engage vendors to help them with that. And I think that’s what we’re going to see for the next probably year or two is a lot of small pilots versus large-scale enterprise initiatives. 

Brendan Hughes: And Leigh, would you agree with that sentiment?

Leigh Sheldon: Yeah, I definitely agree with that. The one thing I would just also offer in terms of that question is in many ways it completely depends on how one is defining AI. As an example, something across industry that we find happening is AI is a very prominent buzzword, but elements of machine learning, natural language processing, the continuum of AI that’s been around for decades, some of that might be happening, but generative AI, agentic AI, the much newer forms of AI, I think that definitely sways and leans exactly with how James answered that question. 

Brendan Hughes: Great. And that’s interesting about the definition around there because I think sometimes that’s where I hear from people where they feel sometimes like they’re behind because they hear such advancements so quickly, but in some cases maybe they really aren’t that far behind. And is that something that you might be focusing on or some of our listeners should start to focus on is what about AI expanding past, maybe like just the traditional automation? Is that something that would beneficial to some courts that are starting to feel a little bit behind? 

Leigh Sheldon: Yeah. One thing I would offer is something that’s really prominent right now is the notion of an AI use case evaluation and prioritization framework. And we focus more often in thinking what’s the problem to be solved and how to enable it. And often what we’re finding is actually traditional automation techniques are equally as value add in solving some of our core environment challenges versus feeling like you might behind because you’re not using AI to solve it versus let’s solve it with what we have today, if that’s right in front of us versus the notion of behind or what’s our level of maturity versus let’s solve the problem and if we can do it today with what we have, that’s a win across the board. 

Brendan Hughes: Yeah, that’s a great point. And I like that idea of the focus on solving the problems for today the best way you can. James, in talking a little bit about this with you prior, I’ve heard something that you touched on and you mentioned it just earlier too, about starting small spot pilots and maybe starting in like a sandbox environment or and starting there as a place where you can start to test and see the validity of some of these things. Is that the case in your mind? 

James Young: Yeah, I mean there’s a lot of different scenarios that I see out there that people are already starting to adapt. Right. I mean, one of the challenges you find is when people come in to Leigh’s point and say, oh, we need to use gen AI to solve our problems. And they try to work backwards from that versus starting with, you know, what are the things that we can do to help automate things or to make life easier on our employees to drive quicker speed to resolution, specifically as it relates to people in criminal situations or civil situations. 

So we see a lot of the small pilots starting to experiment with that, but there’s a lot of them that can be accomplished with say, robotic process automation or predictive modeling for scheduling that don’t require gen AI or agentic AI, not that those don’t have places in the courts. We’ll see Genai really start to expand when it comes to summarizing court findings, to helping judges take their notes and summarize them into findings. We’ll see it used in automating the classification of documents. Things of that nature we’ll start to see happen more and more over the next year or so. 

Brendan Hughes: Great. And then I think one of the things is the next year starts to unfold, and people start to think about things like you were just saying, those use cases a court leader might today be asking, you know, I know that matters. Those sound great. Where do I start, Leigh? What are some of the first things they should start doing? 

Leigh Sheldon: Yeah, first off, kudos to them for even asking that question because that’s step one, having the conversation and thinking through what do we want to achieve, what are the problems we want to solve for now, today, and in the future. And we highly encourage the notion of an actual AI strategy and grounding that because not only do staff have limited capacity, they have full-time jobs. There’s budget aspects that feed into all of this in the tech world. And having a vision of how you’re ultimately looking to mature across the court’s holistic life cycle, that grounds the vision and then that helps inform what do we need in order to enable it. Oftentimes it’s a bit suffocating because there’s so many vendors on the market, there’s so many new tools and technologies that they’re procurement decisions that have to go into this as well. 

What technology do you have today and what do you need in the future? So the only other element I would just highly recommend is the notion of a use case assessment and understanding what are the high reward, high return type of use cases that are going to lead to the best impact for the courts and focusing on prioritizing those, starting with a pilot as an example, we have not seen across any industry to date. Any. Anyone think a big bang approach is the most advantageous way to go about leveraging a new. But getting started with a small pilot, learning through that process before operationalizing it would be ideally the first six months of that journey. 

Brendan Hughes: Yeah, that makes total sense. And even taking a step back from that, I’ve heard others talk about this idea of even around the technology, the governance of it thinking through. I know with courts, you know, security is a big deal and the quality of the information, data. James, is that something that you hear about as well, the data and the governance around some of these projects? 

James Young: Yeah, I think there’s a couple of ways that I’ve seen that come up. One is when you’re using LLM or a large language model that isn’t something curated and developed just for you’re not always sure what’s exactly in there. If it’s also hosted and managed by someone else and you’re feeding in confidential information or personal identifiable information into those, security comes into play there. And then when you’re curating your own data set, you will oftentimes. And a lot of our courts have disparate court systems. So data coming from multiple court case management systems, E-filing systems, E-arraignment systems, all have different data models. And if you don’t curate that data and a common data model, you can have inconsistencies in how your data fits in there, which can also lead to issues with how you leverage that data downstream. 

And putting together that data set, you have to have security around that. So we’re seeing that kind of come up in multiple different ways as people progress through the journey. 

Leigh Sheldon: In many ways, the AI boom has done something great in my mind and put even more focus around the data itself and just how important it is to have quality data, centralized data that makes it much more conducive for more robust analytics or application of AI. But there is the never ending debate of what comes first, perfect data or AI. And so this is where the notion of starting with a small scale pilot with trusted data, thinking about where do you start? We like to look at value, risk and feasibility. When we’re evaluating, you know, what is the most advantageous and someone may find not all of our data is of the highest quality. Quality this piece of it is. And let’s think of what use case we build off of that as an example. 

Brendan Hughes: Yeah, that’s a great point. You know, as you were mentioning a few minutes ago about all the different vendors and I feel even in my space in the marketing world, you know, I get inundated with offers and opportunities for lots of cool tools, AI specific, but trying to manage that all, make it centralized to what you are doing. Just the thought of it gets very overwhelming. So the idea around starting small and then using data that or a subset of data that you feel confident is clean makes a lot of sense. And then the way that this AI technology moves so quickly in that, in your both of your estimations, what do you see in the next 12 to 18 months for the development of these tools, specifically in the justice and court space? 

I know James, you talked about a little bit of the slower adoption is needed. You know, you can’t, you know, just start something to fail and iterate over it. But what’s your sense of where this is going in the next 12 to 18 months? 

James Young: I wouldn’t even hazard a guess at how fast it’s going to evolve. I mean, I don’t think any. Well, I’m sure there are some people that have predicted it, but the speed at which it’s evolving and the use cases are developing just continues to amaze me. So, thinking about where it could be in 12 to 18 months, it’s hard to fathom, especially as it relates to courts. I think what we’ll see is one people get more and more comfortable with what gen AI and agentic AI can do, which will increase user adoption. A lot of the things that you’re seeing now are moving into the more complex decision making and things of that nature. 

I think in order for AI to be more and more adopted or those new features to be adopted, it’s going to be the explainability of the decisions that are made that will really drive the adoption. Some of the stuff we’ve seen in the news over the last year is people will ask AI a question and then they’ll ask the developers to explain how it got to that answer, and they can’t explain how it got to that answer. They’re like, that’s how the neurons formed, you know, and you know, that’s how the neural network is set up. It’s based on the information it’s been trained on. You can’t use that as an explanation when you have a human in a loop validating decisions that are being made, especially as it impacts someone’s livelihood. 

So I think we’ll see more and more tools that are leveraged towards that which will enable more adoption in the public sector in as that moves forward. Leigh, what are your thoughts? You’re probably closer to the ground on this than I am in terms of what’s coming out with the latest tools. 

Leigh Sheldon: Well, I’ll be the first to admit that anytime someone asks, you know, what is your prediction? I, I think there’s a lot of opportunity that I will say, but the technology will continue to advance. But what I foresee more than anything is an expanded application of AI from a productivity lens. So your Gemini, your copilot, your cloud for everyday productivity enablement that is less risky, that is more inherent and natural for staff across a variety of different type of skill sets, and more strategic decisions being made around buy versus build. Notably, if courts are to buy and ProCure a specific AI solution, they’re ideally thinking of it from a holistic lens. So I think that’s where a lot of time and effort will be coming from. 

Leigh Sheldon: Courts in particular, with one additional facet which is the focus on to enable that technology, what do we need to do about our data? So I think it’s going to draw a lot more generation of discussions around the data itself as well over that timeframe. 

Brendan Hughes: Yeah, it’s interesting that it keeps coming back to the data, which is something that’s been around forever, is you have data and information and it’s how do we access it, use it to make better decisions. And even though we’re talking about these very creative and amazing tools, it still could come back to how are you managing your data? Is it good or is it bad or what are you doing with it? So we’ve touched on a number of things here, and if there was, I’m going to ask each of you if there was a couple of things that you would want a court or a justice leader who’s listening to this to kind of take away from our conversation, you know, about where things stand in AI, where they should be focusing their efforts. 

What are those one or two things you’d like them to take away? 

James Young: Not to beat the horse, but the data is. I would put kind of two things equally. One, you should have a data strategy, because I think in our research only five states have statewide single court case management systems. And even then they have multiple E filing systems. So there’s always multiple sources of data and not all of them have standardized common data models where all of that data feeds up to. So having that data assessment and strategy would be number one, and then number two, starting with your governance and your policies around AI. So before you even start thinking about the pilots, come up with a framework for how you’re going to evaluate them, how you’re going to monitor them, how you’re going to move them through the process. And then, you know, starting with small pilots. 

And Leigh and her team do a lot of work with this across the country, helping state agencies to come up with their policies and procedures, as well as to set the. The common data models and data. We’ve actually been working quite a bit to kind of standardize a common data model for Quartz just so that we can show people basically how you start that journey and how you can build dashboards off of that and then how you can leverage that to create specific use cases for AI. 

Brendan Hughes: Great. And Leigh, what do you want people to take away from today? 

Leigh Sheldon: I would want folks to take away more of the sentiment that there is no standard across the board in terms of where you have to be at today in terms of your AI and data journey. Those that have even just started the conversation are moving in the right direction. It always starts with the conversation and thinking through what are we going to do, how do we make sense of this and what do we want our priorities to be? Every state court justice will be unique in terms of what they might prioritize, and neither is better or worse than the next. And then the only other thing just to echo James’s sentiment is there is a lot to think through, but you’ll only be able to move forward if you have the policies and standards in place that allow you to do so. 

Leigh Sheldon: Whether that’s from the pure court’s judiciary lens, the state lens, but also other state pertinent laws and regulations that you would need to adhere to. 

Brendan Hughes: Great. Yeah, I really appreciate that. The first part too, Leigh, where kind of going back to the top where some people might be feeling behind or overwhelmed or both overwhelmed and behind. But it’s nice to know that wherever you are at that journey, you’re really not behind. That’s where your journey is. And there’s places to go from here. And I believe that today that both of you guys gave us a lot of great insight. So I appreciate that court leaders and justice leaders can take with them to start their journey or to continue that journey in the best way. Thank you, James and Leigh again for joining us today. And until next time, this has been Scaling Justice.