By Sam Glover
AI-powered online legal tools fill a critical need. According to the Legal Services Corporation’s justice gap study, LSC-funded organizations turn away at least half the requests they receive due to limited resources (in Boston, over 60 percent must be turned away). Potential clients who are turned away must try to solve their legal problems on their own, while legal aid organizations face growing need and shrinking resources. (Maybe that’s why legal aid organizations are adopting AI solutions at twice the rate of the rest of the profession.)
At the same time, most Americans do not even seek help from a lawyer for their civil legal problems. Most people with a civil legal problem try to solve it on their own. That means they must navigate a legal system that is designed by and for expert users—lawyers—not regular people with everyday legal problems. This puts them at a serious disadvantage.
There are many ways AI technology can help reduce the disadvantages experienced by those who are not represented, including:
- Guiding people through completing and filing court forms and other legal documents
- Streamlining routine processes so legal aid organizations can serve more clients
- Helping self-represented litigants access and understand the law
We’re building some of those tools—carefully and responsibly—at Suffolk Law’s Legal Innovation and Technology Lab, in collaboration with courts and legal aid organizations across the country. So are other law schools and entrepreneurs in and around Boston and beyond. In this article we will examine several tools that leverage artificial intelligence to increase access to justice:
- The LIT Lab’s Court Forms Online guided interviews for court forms
- Lemma’s AI-assisted intake workflowsfor legal aid
- Descrybe’s simplified case law summaries
Will AI replace lawyers?
But first, let’s briefly address the elephant in the room: will AI-powered tools replace lawyers? Well, maybe someday, but not yet.
Lawyers have been preoccupied with losing their monopoly at least since the invention of the printing press, but there are more lawyers than ever, working more than ever, and there still aren’t enough lawyers to close the access-to-justice gap.
AI’s “jagged technological frontier” means AI is better than a human at some tasks and worse at others. Similarly, when a human works with AI, together they are better than a human or AI alone at some tasks, but worse than humans or AI alone at others. In other words, sometimes AI is better, sometimes a human is better, sometimes they are better together, and sometimes AI makes both worse. According to the authors of the paper that introduced the jagged frontier concept, “It is clear that the best approaches to using AI are not fully understood and need to be deeply examined by scholars and practitioners.”
To help with that examination, Dazza Greenwood, an MIT professor, Boston-based entrepreneur, and Suffolk Law graduate, developed Lake Merritt, an open-source tool intended to help lawyers evaluate AI quality. A lawyer first defines the tasks (prompts) and the results they expect for each task in a spreadsheet or structured text file. Then, Lake Merritt gives the tasks to the AI to be tested and collects the results. A second AI “judge” scores the results by comparing them to the lawyer’s expected results. Reports contain a detailed analysis with the AI judge’s reasoning for evaluating the evaluation. Lake Merritt is free and open-source, but Greenwood is also offering early-access evaluations for a fee.
AI is a powerful and sophisticated tool, but it is still a tool, not a panacea—or a replacement for a lawyer. When used carefully and responsibly, AI can increase the effectiveness of online legal tools, and it offers some new ways to solve old problems. Americans with legal problems are desperately in need of access-to-justice gap fillers, and AI can help meet that need.
Here are three examples that show how.
Online guided interviews
The Suffolk LIT Lab’s Court Forms Online website has dozens of online guided interviews to help self-represented litigants complete and file court forms. These guided interviews embed legal knowledge into a rules-based expert system to help users complete and file complex legal forms successfully. The online guided interviews on Court Forms Online are designed to be used from a phone or computer. Some, like the eviction sealing petition launched this past spring, can be e-filed directly with Massachusetts courts.
Many of the guided interviews on Court Forms Online are built by law students in Suffolk Law’s Legal Innovation and Technology Clinic, taught by David Colarusso and Quinten Steenhuis (with some help from me). In the LIT Clinic, students collaborate with the Massachusetts Appeals Court and Trial Court, as well as other courts and legal aid organizations. Students build, update, and maintain online guided interviews using open-source software, including Docassemble and the LIT Lab’s Assembly Line tools. The LIT Clinic (and similar clinics, courses, and labs at other law schools) trains law students to design and build technology solutions to legal problems. They build tools, test prototypes, and evaluate how technology intersects with ethics and professional responsibility. This results in lawyers who understand and have experience with the potential and pitfalls of technology—including AI—and have a robust toolbox for solving legal problems.
Court Forms Online is part of the LIT Lab’s Document Assembly Line project, a collection of open-source tools for court forms, guided interviews, and e-filing. Through the Document Assembly Line, the LIT Lab supports and provides hosting and e-filing services to a growing community of courts and legal aid organizations across the country that are also building online guided interviews.
Streamlining legal aid intake
Since legal aid organizations receive more than twice the requests they can handle, it requires significant staff time to interview and qualify people who ask for help. To lighten that load, Boston’s Lemma builds AI-powered intake workflows for legal aid organizations, including Greater Boston Legal Services, the Virginia Legal Aid Society, Mid-Missouri Legal Services, and the Oregon State Bar’s lawyer referral service.
Lemma customizes its intake workflows for each client, but AI components can include:
- Voice conversations, where the AI speaks with the potential client, asking questions, listening to answers, and following up
- Forms that start the process, then AI-created follow-up questions based on the potential client’s entries
- Text conversations, or AI chatbots
- AI-driven follow-up, in which the AI can automatically generate and send questionnaires to gather missing information or trigger follow-up activities, via email or SMS
Humans are kept in the loop, especially for emergencies, safety issues, and ADA accommodations. The potential client is always informed when engaging with AI, and humans make all final decisions.
By taking on some of the more time-consuming steps of the intake process, AI-assisted workflows free up precious legal aid resources so organizations like Greater Boston Legal Services can serve more clients.
Simplifying case law
Case law is complex, but Descrybe uses AI to simplify it. Descrybe was created by former Suffolk Law communications director, Kara Peterson, and her husband, Richard DiBona, in Greater Boston. It offers free plain-language summaries of U.S. case law in English and Spanish, plus simplified summaries in both languages, calibrated to a 5th-grade reading level. (Here is Marbury v. Madison, as one example.)
Descrybe also offers an AI-powered legal research toolkit for a fee. Paid tools include the Cytationator brief checker, which analyzes citations to determine whether they actually support the underlying proposition. Other tools include a brief checker, legal issue explorer, and instant translations.
It’s relatively easy to find free case law, but that mostly helps people who are already used to interpreting opinions. Descrybe’s AI-generated plain-language summaries and translations make case law easier to understand.
The near future
Those are just three local examples. Academics, practitioners, companies, and courts are all looking for ways to address increasing need (or decreasing legal budgets in the private sector) with AI.
Here are some innovations you can expect to see in the near future:
- AI document analysis to help self-represented litigants understand legal notices and documents and guide them to the resources they may need to respond
- AI drafting assistants that can help self-represented litigants and volunteer lawyers create more persuasive, complete court forms and other documents
- AI-assisted triage for legal aid organizations to help route clients to the resources or referrals they need
- AI filing assistants that can check documents for completeness, formatting, and required attachments before submission
- Judicial workflow assistants to automate routine tasks, summarize filings, or screen for procedural compliance
All of these are already in various stages of formal and informal testing right now by academics, practitioners, companies, and courts.
Online legal tools, used responsibly, are an access-to-justice multiplier
The tools described above represent different but complementary strategies: making court forms and legal documents easier to complete and file, making legal aid organizations more efficient, and making the law easier to understand. Each addresses a different friction point, where those without legal training and experience are likely to be excluded or overwhelmed.
With any innovation, the challenge is to find its jagged frontier—the tasks where the technology can improve on existing results. That’s why when AI is used in online legal tools, it is almost always used for specific, narrow tasks where the AI is proven to be effective.
Since most people with a civil legal problem don’t have a lawyer to help them, it’s not appropriate to compare the results of an online legal tool with help from a competent lawyer. For an online guided interview, an appropriate comparison might be to assess whether a self-represented litigant is more likely to complete and file a court form successfully (perhaps using a cell phone) with AI assistance than if they have a bare PDF form. For an intake workflow, it might be appropriate to compare the mistakes made with an AI-powered workflow to a human alone.
The overarching goal for access to justice, of course, is to make it more likely that someone with a legal problem will get the help they need, whether that means finding, completing, and filing a court form, securing representation by a legal aid lawyer, or advocating effectively for themselves. Each improvement, from the printing press to standardized court forms to online guided interviews to AI assistants, multiplies the impact and chips away at the access-to-justice gap.
Sam Glover is a Clinical Fellow at Suffolk Law’s Legal Innovation and Technology Lab (LIT Lab). Sam is an American Bar Association Legal Rebels Trailblazer, a Fastcase 50 honoree, and writes for the LIT Blog as well as his own.
