While law firms and other private entities have so far been at the forefront of the legal industry's experimentation with generative artificial intelligence, experts say that court systems can play a role in deploying the technology to help self-represented litigants navigate court systems, resolve disputes remotely, and fill out required forms.

Several courts are already using automation and AI for website chatbots, online dispute resolution and form completion to help pro se litigants, and adding generative AI can expand these initiatives, legal experts say.

Bridget McCormack, president and chief executive officer at the American Arbitration Association and former chief justice of the Michigan Supreme Court, recently told Law360 that generative AI can help courts provide better services to members of their communities who may lack the resources to hire an attorney to assist them with their legal problems.

And by courts taking a more proactive role in stepping up to help, she said, it could also help increase the public's trust in them.

"Public confidence in courts has been on the decline for some time, and I personally think that the number of people navigating justice problems without lawyers, who don't understand how to help themselves … is contributing to the decline in public confidence about courts and courts' work," McCormack said.

She added, "If, on the other hand, they all of a sudden had access to the information they need to be able to navigate some solutions to those civil justice problems, and they knew that that help and information was coming from the courts — what an enormous gain for public confidence in courts and what they do."

Making Navigating Court Info Easier

Several state courts already have chatbots on their websites that use a form of AI to make it easier for litigants to find relevant information, according to a report released in January by the National Center for State Courts. These chatbots allow users to ask questions, then direct them to relevant information and resources.

For example, the Eleventh Judicial Circuit of Florida in Miami launched an AI chatbot called SANDI in 2022 that can answer written or spoken questions in English or Spanish. Users also can select options from a menu.

Implementing a chatbot is an easier and less expensive way for courts to make their websites easier to navigate and understand than revamping their entire websites, according to NCSC's report.

Miriam Kim, partner at Munger Tolles & Olson LLP and fellow at the Berkeley Center for Law and Technology, said that generative AI like OpenAI's generative pre-trained transformer, or GPT, models can improve court chatbots by expanding their language understanding capabilities.

She noted that most existing court chatbots require users to word their questions in a certain way to trigger a programmed response.

"Using a large language model like GPT, the chatbot can really understand a person much better and direct them to the right section of the website," she said.

Courts can also use generative AI to create chatbots to help users with specific legal issues like evictions or expungement, according to a research paper co-authored by Kim.

Kim and her co-authors said in their research paper that they used OpenAI's GPT builder and information from Arizona state court websites to create chatbots for expungement and evictions in Arizona.

The expungement chatbot can determine whether a user is eligible for expungement based on Arizona law and explain the process, according to the paper. The eviction chatbot can answer questions about failure to pay rent and lease violations, and explain options available to tenants and landlords. On the other hand, the paper noted that sometimes the chatbot provided inaccurate information.

"Despite the present limitations of LLMs, the Arizona eviction bot illustrates the potential ability to train LLM-based tools to make self-help information on court websites more accessible to pro se litigants and members of the public," Kim and her co-authors said in their paper.

Resolving Disputes Remotely

Several state courts offer online dispute resolution that allows parties to resolve some types of disputes like family law or housing matters outside a courtroom, according to a report released by the American Bar Association's Center for Innovation in 2020. States that have online dispute resolution include Michigan, Ohio, Arkansas, Texas, New Mexico, Arizona, Utah and California.

While McCormack was chief justice of the Michigan Supreme Court, she helped the state judiciary roll out an online dispute resolution system called MI-Resolve.

The online system allows disputing parties to have a text-based conversation along with a trained mediator, who tries to come up with options for resolving the dispute, according to the Michigan state court system's website.

The types of matters that can be addressed through MI-Resolve include custody arrangements, contract cancellations, professional services billing and post-termination wage collection claims, and landlord-tenant disputes over rental deposits or alleged property damages.

McCormack said that MI-Resolve was launched in a few counties in 2019 to test the system and get feedback, but after the COVID-19 pandemic started in 2020, MI-Resolve was made available in all of Michigan's 83 counties.

McCormack said that generative AI can be incorporated into online dispute resolution tools like MI-Resolve by training large language models to handle and resolve specific types of disputes, similar to the way that Amazon and eBay automate the dispute resolution process for customer complaints on their platforms.

Or generative AI could be used to automate the process of identifying where parties can't agree, clearing the way for the dispute to be sent to a mediator to devise a solution, according to McCormack.

"Those kinds of systems would be a huge advantage to courts in navigating some of their high volume dockets," McCormack said.

Streamlining Form Completion

Some of the states that use document automation software to help self-represented litigants fill out court forms include New York, Illinois, Michigan, Florida, Texas, Pennsylvania and Massachusetts, according to legal document assembly website LawHelp Interactive.

Automated document assembly tools work by guiding litigants through a series of questions and then pulling all the answers into a standardized form that can be filed in court.

Automated forms are available for a wide range of legal issues including divorce, child custody, evictions and foreclosures, and for wills, according to LawHelp Interactive.

For example, the New York state court system has more than 20 do-it-yourself form programs for legal matters such as uncontested divorces, custody or visitation modifications, guardianships, name changes, consumer debt, foreclosures and evictions.

Where generative AI could take court-provided assistance to the next level, Kim said, would be applying it to forms where pro se litigants have to write a narrative.

She noted that writing a narrative for forms like a request for a restraining order can be difficult for some self-represented litigants, especially if English is not their first language.

"A [generative AI] tool can help pro se litigants find the words they need to explain, whether it be just because their English is limited or because they don't know the legalese that they need to say," she said.

Overcoming Generative AI Challenges

While tools like these can help pro se litigants more easily navigate legal issues, courts face challenges when implementing generative AI, including lack of funding and in-house technological talent, data security and privacy concerns, and the tendency for generative AI models to output false information.

Zach Zarnow, deputy managing director of NCSC's access to justice team, said that courts can address their concerns about data security and privacy in vendor contracts.

"Being very thoughtful about your vendor relationships and contracting and procurement can be really important to making sure that at the very least, everybody understands the respective roles and responsibilities and that the data ownership and usage is defined," he said.

McCormack added that courts could try to overcome their limited financial and human resources by partnering with academic institutions and nonprofits to build new services, tools and products.

For example, Georgetown University Law Center in March announced the creation of a fellowship to give state courts an opportunity to work with technologists and software designers to develop tech-based solutions to improve access to the judicial system.

"It's a great opportunity for courts to approach it across the country the way they did responding to the pandemic, like, let's pool our resources, pool our chief technology officers, pull in some academic support, pull in some nonprofit support and see what new things we can build right now, and then when they work, replicate them," McCormack said.

--Editing by Robert Rudinger.

Have a story idea for Access to Justice? Reach us at accesstojustice@law360.com.