Judge Samuel A. Thumma
Generative artificial intelligence is all the rage these days. People involved in the practice of law are fascinated by generative AI — looking at what it can do better and faster in the practice; the opportunities and dangers it poses; ethical issues; the unauthorized practice of law, which is a crime in many places; and even whether it will make lawyers or judges obsolete.
As a judge who will retire from the bench in less than a year, I am pretty confident generative AI will not replace me. But more broadly, I am excited about the potential of generative AI and the law, particularly in how it may enhance access to justice efforts. And, as a judge, I have thought a fair bit on how courts can use generative AI to enhance access to justice systemically.
For the past dozen years, the National Center for State Courts has conducted an annual "State of the State Courts" survey. The 2025 survey results, published on Dec. 10, show the need for courts to do more in using generative AI, and in showing the public that generative AI can improve the judicial system.[1]
The respondents included a representative sample of 1,000 registered voters. When asked which of two statements about AI came closer to their views, 51% of the respondents selected "AI will hurt state courts by increasing the risk of mistakes that judges and staff can't always catch and making it harder for people to trust court decisions."
Strong medicine. But 31% selected "AI will help state courts work faster and more efficiently, by cutting down on paperwork, reducing backlogs, and giving judges better tools to manage cases."
The remaining 18% were unsure.
The NCSC correctly observed that the results present "courts with both practical and [public relations] challenges as they move to adopt AI responsibly." More bluntly, the report concluded, "Americans bring a 'glass half empty' perspective to how AI will impact state courts."
What, then, are courts to do with generative AI to enhance access to justice and, along the way, engender confidence in use of technology in the court systems?
The basic and obvious answer, to me, is to do something — actually, a whole bunch of somethings. Generative AI is not going away.
We in the judiciary are a tradition-based crew, at times for very good reasons — think of following precedent and the rule of law. But the COVID-19 pandemic forced courts to change, looking at technology in new ways, with much good as a result.
In May 2023, summarizing lessons learned from COVID-19, and as generative AI was becoming a phenomenon, I wrote that our pandemic experiences provided a once-in-a-lifetime chance to improve access to justice.[2] Now, generative AI presents a second once-in-a-lifetime opportunity to improve access to justice.
What, then, should courts do to seize this opportunity? A whole bunch of somethings that I will put into five categories of suggestions here.
First, courts should learn about generative AI: the basics of how it works, what it is and is not, how it is evolving, the opportunities and risks involved, issues it raises for ethical and other restrictions applicable to lawyers and judges, and other facets of generative AI in the courts.
I am not suggesting that judges need to be computer scientists. With no false modesty, I am the least technology savvy person in the room — any room, including my dining room at home. But having a basic understanding of these aspects of generative AI is important. To me, a basic resource I find helpful — recognizing it is almost two years old now, and I had a hand in preparing it — is a Joint Technology Committee resource bulletin titled "Introduction to AI for Courts."[3]
Second, judges should try using generative AI. That sounds basic, I know. And I also know that using popular internet search platforms these days involves generative AI use even without the user necessarily knowing it. But I have been surprised by how many judges I meet who say they have never used generative AI.
In the past two years, I have had a hand in judicial education programs on generative AI efforts in maybe half a dozen state judiciaries, and some national and international efforts as well. Each time, when beginning the presentation, I ask for a show of hands about how many people have even tried generative AI. I have never had more than half the hands go up in response — indicating that at least half those in attendance have never knowingly used generative AI.
Although a bit surprising, that type of response presents a wonderful opportunity for growth. The reasons for not even trying generative AI undoubtedly vary, ranging from fear to angst, tradition and other explanations. To try to break down those barriers, when we can, I offer up a live generative AI demonstration, asking for questions from the audience that we run in a generative AI platform in real time.
Before doing so, I promise the group three things: (1) The FBI won't come busting in to raid the place, (2) court security officers won't come into the room to ask what's going on, and (3) the sun will rise in the east the next morning. So far, my promises have stood. And the responses from the judges in the room when they see, for the first time, the power of generative AI is really rewarding — even if, as I typically suggest, our first generative AI search is for the best risotto recipe.
Third, courts should develop use policies and provide training for appropriate generative AI use within the court. More recently, when I speak on generative AI to judicial groups, I ask how many people know if their court has a generative AI use policy or have had generative AI training. Maybe one in 10 hands go up in response.
To me, using generative AI, and having training and use policies on generative AI, go hand in hand for responsible and beneficial use of it in the courts.[4]
Fourth, judges and those with whom they work should get their hands dirty using generative AI to see how it works in the legal context. There are many ways that can be done.
One way, for sure, is to work with really smart people who know both technology and the law. But beware: There are more people who claim to be generative AI experts than there are actual generative AI experts, and care should be taken, particularly in dealing with vendors.[5]
The NCSC also has an AI sandbox offering what it describes as "a secure, hands-on environment for judges and court professionals to familiarize themselves with AI tools."[6] Again, I'm no technologist, but this offers an important opportunity to get one's hands dirty by using generative AI tools in a safe environment.
When I speak with judges about generative AI use, I encourage them to try it. If they do and they decide not to use generative AI, they will have made an informed choice. The AI sandbox is a way to try out generative AI in a safe place, and then make that informed choice.
Fifth, and finally, judges and those with whom they work should think hard and creatively about how they can adopt generative AI to enhance access to justice. An inspirational way to do that is to look at what other courts are doing that could be emulated in their local court. Let me offer a few examples.
Marion County, Indiana, is doing incredible things with generative AI in a pilot program to expedite time-sensitive mental health appeals.[7]
Various courts are using generative AI-driven virtual assistants and chatbots — not the old, pre-generative AI chatbots that I hated because they seemed to always go in circles and never provide much help, but newfangled tools that provide answers, are helpful, and learn and do better with time.
The Nevada Supreme Court has an AI chatbot that offers legal guidance in multiple languages.[8] As another example, Florida's 11th Judicial Circuit has a chatbot named SANDI — an acronym for Self-Help Assistant Navigator for Digital Interactions — to assist litigants with navigating the court system.[9]
This is not an exhaustive list. In the summer of 2024, I shared some thoughts in Judicature about other ways courts can use generative AI to further access to justice.[10] The NCSC has also identified other specific generative AI uses for courts to enhance access to justice, including document handling, workflow management, data processing, court management and litigant portals.[11] And it is important to remember that we can learn from what courts in other countries are doing to use generative AI to further access to justice.[12]
There are an almost limitless number of creative generative AI uses courts can identify and implement to enhance access to justice in the judiciary. In my mind, a desire to do things better, in light of specific needs, can drive those advancements.
I also think that a small team (in my mind, three people: an IT person, a court administrator and a judicial officer) who are committed to doing so can make an enormous difference in innovatively using generative AI to advance access to justice in the court.
I look forward to seeing how courts use generative AI to enhance access to justice in the future, and am optimistic in those endeavors. For now, I hope that these five categories of suggestions, capturing a whole bunch of somethings, can help empower and prompt action by those wanting to use generative AI in the courts to enhance access to justice.
Using generative AI in the courts is a second once-in-a-lifetime opportunity to improve access to justice, and the time to do so is now.
Samuel A. Thumma is a judge on the Arizona Court of Appeals, Division One. He previously served as a judge on the Arizona Superior Court in Maricopa County.
The opinions expressed are those of the author(s) and do not necessarily reflect the views of their employer, its clients, or Portfolio Media Inc., or any of its or their respective affiliates. This article is for general information purposes and is not intended to be and should not be taken as legal advice.
[1] NCSC 2025 State of the State Courts - NCSC 2025 State of the State Courts, https://www.ncsc.org/resources-courts/state-state-courts-2025-public-opinion-poll-findings.
[2] "We have a once-in-a-lifetime opportunity to improve access to justice – let's not squander it," The Hill (May 24, 2023), https://thehill.com/opinion/judiciary/4016923-we-have-a-once-in-a-lifetime-opportunity-to-improve-access-to-justice-lets-not-squander-it/.
[3] https://ncsc.contentdm.oclc.org/digital/collection/tech/id/1191/rec/2.
[4] Examples of such court policies include Arizona's, Arizona Code of Judicial Administration Section 1-509 (Use of Generative Artificial Intelligence Technology and Large Language Models), California, https://newsroom.courts.ca.gov/sites/default/files/newsroom/2025-02/25-044%20JC%20Model%20Policy%20for%20Use%20of%20Generative%20AI_PPT_FINAL.pdf, Illinois, https://www.illinoiscourts.gov/News/1485/Illinois-Supreme-Court-Announces-Policy-on-Artificial-Intelligence/news-detail/#:~:text=The%20Illinois%20Courts%20will%20be,making%20will%20not%20be%20tolerated, and South Dakota, https://nationalcenterforstatecourts.app.box.com/s/9c9c6h90v4bo9qv06r82r63f1g0zw2ek. See generally https://www.ncsctableauserver.org/t/Communications/views/AIRRT/AIResources?%3Aembed=y&%3AisGuestRedirectFromVizportal=y ("Sate court guidelines & more).
[5] Among others, I have had the great fortune of working with Professor Cas Laskowski, Associate Librarian and Head of Research, Data & Instruction (and other things) at the James E. Rogers College of Law. She has helped me understand generative AI far more deeply than I could by reading about it and has also used various generative AI platforms in my efforts – choppy and incomplete as they are – to try to make a court rule set often used by self-represented parties more readable and easier to understand.
[6] https://www.ncsc.org/resources-courts/ai-sandbox.
[7] https://www.in.gov/courts/files/order-other-2024-24S-MS-190.pdf; See also Greg Pachmayr, "Involuntary Commitments, Artificial Intelligence & Expedited Appeals" (May 5, 2025), https://times.courts.in.gov/2025/05/05/involuntary-commitments-artificial-intelligence-expedited-appeals/.
[8] https://app-backend-brnngeqazsnem.azurewebsites.net/.
[9] https://www.jud11.flcourts.org/.
[10] "How to Harness AI for Justice: A Preliminary Agenda for using Generative AI to improve Access to Justice," 108 Judicature 42, https://judicature.duke.edu/wp-content/uploads/sites/3/2024/07/AIforJustice_Vol108No1.pdf.
[11] https://www.ncsc.org/resources-courts/leveraging-ai-reshape-future-courts#:~:text=Document%20handling:%20AI%20can%20automate,based%20on%20expertise%20and%20experience. (Dec. 9, 2025).
[12] "AI in justice administration and access to justice," Organisation for Economic Co-operation and Development (Sept. 18, 2025) (overview of efforts worldwide), https://www.oecd.org/en/publications/2025/06/governing-with-artificial-intelligence_398fa287/full-report/ai-in-justice-administration-and-access-to-justice_f0cbe651.html.
For a reprint of this article, please contact reprints@law360.com.
