How AI is helping and hurting Arizona courts

Artificial intelligence is affecting lots of areas of life, both public and private. And that includes the judicial system.

Early last year, Arizona's judicial branch formed a steering committee on AI to figure out how the technology should — and maybe should not — be used in the courts.

Dave Byers is director of the Administrative Office of the Supreme Court.

The Show spoke with Byers about how the judicial system is implementing AI, starting with whether there’s an overarching philosophy on AI there.

Arizona Supreme Court’s Administrative Office of the Courts

Full conversation

DAVE BYERS: Well, you know, back, a couple of years ago, we held the first national conference in Arizona on the use of AI and in the law and the courts. And that kind of led us down the path.

So we're, we're interested in exploring the potential uses in, in a whole variety of different ways.

MARK BRODIE: What has been successful so far for the courts?  Like, what have you tried and said, OK, this is, this seems good. 

BYERS: Yeah, well, the first thing which has gotten a lot of notoriety is our AI reporters. So, so what the Supreme Court issues an opinion like they did today, they, they can be long and rather complicated. They're, the court's writing for attorneys and, and so people understand why they're making the decision they're making, but they could be complicated for reporters and complicated for the public.

So what we've done is, we've created two avatars, we call them Daniel and Victoria, and they do a 1- to 2-minute video summary of the opinion using an AI avatar. That appeals to folks that want to get a quick summary of what the opinion means. That's been highly successful. It's got worldwide attention. And it's driven our, people coming to our website to look these up.

BRODIE: And to be clear, these are not AI like they're not generating the content, right? Like the justices are, are writing the scripts and just the AI avatars that are reading them. 

BYERS: That's right. The scripts and the, and the reporter says that right off the bat, that the script is written by the justices and to summarize the case, but they're, they're the, they're reporting to the public. And the reason we do that, you know, we're not professionals, we don't have folks that can, can just, professionally, humans do it on a camera, we have to do multiple takes, we have to have people available to have production with the, with the avatars, we can produce these things in just minutes at a very inexpensive cost.

BRODIE: Are you looking to do anything else kind of like that, either for lower courts or for other judicial services? 

BYERS: Well, we have another, another feature coming. We, we also at the court produce a lot of different reports. So for example, we, we held a summit on juvenile violent crime recently and a report came out of that. And we're now using AI software to turn those reports, which we published, but we're gonna turn those reports into podcasts.

So those are two things we're doing to get information out to the public in different types of formats, both written as well as a video.

BRODIE: Are there things that you and the justices and maybe others in the court just are not ready or even maybe willing to use AI for? Like, are there things that it maybe just wouldn't be appropriate in the, in the judicial system, the court system to use AI for? 

BYERS: Yeah, we're never gonna use AI to actually resolve a dispute. We're not gonna replace the judge's judgment with AI. We do provide judges with an AI tool that makes them much more efficient in doing their legal research and in checking pleadings that lawyers or citizens have filed to make sure the pleadings are correct.

There you probably hear these stories about some AI products making up cases and sightings. So we, we have tools for judges to use to make them much more efficient in doing their legal research, but we're not going to replace a judge's, a human's judgment with AI.

BRODIE: And safe to say the justices are not using chat GPT to write their opinions. 

BYERS: No, they, they don't use ChatGPT to write their opinions.

BRODIE: Let me ask you about an incident that happened in a court in Arizona not that long ago, where the victim of a crime, their family basically used an AI-generated script and avatar to let that person address the court. 

Do you foresee more of that kind of thing of, you know, AI being used in new and different ways, if not from the judges and from the courts themselves, but from people who are involved in court proceedings? 

BYERS: Well, I'll tell you, we didn't see that particular instance coming, so it was pretty dramatic and as you said, the, the victim was shot and killed in a road rage incident. And the victim's sister who was making a victim statement, the person was already found guilty, and now the victims have a chance to present a statement in, the law says in any way they want.

She used her brother's pictures and some videos to create an avatar of her brother who then read the script that she generated speaking as she thought her brother was. That was pretty dramatic and unexpected and it's on appeal, so that's gonna go to the Court of Appeals and likely to the Supreme Court and we'll, we'll get some guidance from the appellate courts on whether that's a legitimate use or not.

But to answer your question directly, yeah, we're, we're, we think we're gonna see lots of different ways.

We're, we're seeing it, we're also seeing it with self-represented litigants, people who come into Superior Court now in civil cases and represent themselves. They're generating pleadings and motions using AI. In fact, it's becoming a problem because they can generate so many motions and sometimes repetitive ones. It's becoming a real workload problem for the judges.

BRODIE: Well, I was going to ask if this is the kind of thing where, you know, we hear often about how technology is ahead of the policy that, you know, lawmakers, policymakers can't quite keep up with new technology and ways to make sure it's used appropriately or not. 

It sounds like maybe in some ways that's kind of the situation in which the courts are finding themselves as it relates to AI. 

BYERS: Yeah, that's true. Legislative bodies, particularly, I have a real hard time keeping up with the new technology. To help do that, we created a courts AI committee. And it's, it's put out our governance, so it, it, it tells our own court employees what they can do and can't do with AI and it, it recommends the Supreme Court rule changes or adjustments they need to make as AI issues come up.

That group is maybe more nimble and the courts can react relatively quickly, but we might have a hard time keeping up also.

You know, the other area we're concerned about is people faking evidence, and we've always had people bringing fraudulent documents and things to court, but there's some areas like say in traffic cases, say a person is cited for running a stop sign. Well, it is so easy now with various AI products to take a picture of that intersection, put a circle around the stop sign and say remove it, and then come into court and show the judge, judge, there's no stop sign at this intersection.

And we're, we're anticipating that fake video, deep fake videos, audios and protection, in the orders of protection cases, are gonna come into the court system, and we're having to find ways to deal with that now.

BRODIE: Yeah, how do you try to deal with that? What do you do about that? 

BYERS: Well, we're working with various national groups, both some coming out of sophisticated government operations as well as private sector, to help give some tools to our judges that could help determine a deep fake or fake evidence, fake videos, but it's gonna be a challenge.