Courts were built for lawyers, and for decades, they have worked to close the gap between what their processes require and what people navigating the system without an attorney can do on their own. It has been, in the words of the Hon. Jennifer Bailey, a retired Florida judge, a "huge stressor for judges and courts" that no single strategy, such as plain language forms or self-help centers, has fully solved.

AI presents an unprecedented opportunity to change that. Judge Bailey and Zach Zarnow, Executive Director of Scale Justice, spoke on this issue at a recent webinar, The Coming Wave: How Courts Can Prepare for AI-Assisted Litigation, hosted by the AI Policy Consortium — a joint effort by the National Center for State Courts (NCSC) and the Thomson Reuters Institute (TRI). Their concensus is that the possible change that AI can bring in this area deserves to be treated as an opportunity rather than a threat.

An opportunity, with a catch

"AI presents the first opportunity in my lifetime as part of the justice system to improve access to justice at scale," Judge Bailey says. "But — and here's the but — if you expand access to justice, you expand the demand for delivery of justice."

That is the real story here. Courts have spent generations trying, with limited success, to help self-represented litigants clear the basic hurdles of stating a claim, organizing evidence, and understanding their options. The tradeoff, however, is volume. More usable filings means more cases that are ready to be heard with limited pre-trial intervention — and courts need to be ready to handle that flow.

Judge the filing, not the tool

The instinct, when something new and slightly unfamiliar — like an AI-assisted document — appears in a stack of filings, is to flag it, question it, and treat it differently. However, courts do not do this with any other tool a litigant might use.

"We didn't interrogate whether they use the right volume in the law library 20 or 30 years ago," Zarnow says. "And we don't care whether they're using Thomson Reuters this, or Westlaw that, or Lexis this... What we care about is what's filed with the court and what we do as a court to make sure that we're giving everyone access to justice."

Judge Bailey further refined the same point, noting that what a court needs to know about a complaint has never changed: "Does the complaint state a cause of action? Do we understand the facts from the complaint?" she asks. "That's all we need to know. I don't care if AI wrote it or you wrote it with fairy dust, right? If it does [the job] properly, that's the important thing."

Treating AI-assisted filings as inherently suspect is functionally the same as questioning whether a filing was typed on a computer instead of a typewriter, or drafted using Microsoft Word instead of longhand. It may feel like it levels the playing field or reduces risk, but in practice it just adds friction for the very people who most need a way in to the justice system. The standard for a filing should be whether it meets the requirements of the case, not what tool was used to produced it.

Yes, there are risks & we already knew that

It would be dishonest to pretend that AI use carries no risk, and no one on the webinar tried to claim that is the case. Hallucinated citations, overconfident advice, and filings that look polished but skip essential facts are real problems, and they've already been well documented and widely discussed. That conversation has been happening loudly for a couple of years now.

The point is not that the risks don't exist; rather it's that we, collectively — as courts, as a profession, and as an ecosystem of legal service providers — need to stop treating those risks as the whole story and start putting comparable energy into capturing the benefits. It is easy to get fixated on a new tool, to treat it as the shiny object worth all the scrutiny. What gets lost when we do that is the human in the loop — the clerk, the judge, or the reviewer — whose job is still to check the work, and the human affected by the loop's output — the litigant whose case does or does not get heard.

Treating AI-assisted filings as inherently suspect is functionally the same as questioning whether a filing was typed on a computer instead of a typewriter, or drafted using Microsoft Word instead of longhand.

AI is an innovation, but it is not the goal. The goal has always been a process and a set of outcomes that work better for litigants. Measured against that goal, a tool that helps more litigants appear before a court with a complete, coherent filing is worth building around, not bracing against. However, courts need to keep that human in the loop, one who is checking the output and always ensuring the litigant’s rights are paramount.

Of course, none of this means courts should sit back and let outside tools be the only source of help. Zarnow was direct about the cost of ceding that ground. "If courts aren't trying to win that battle for attention, they are, by default, ceding some ground that I think is really valuable and ultimately causes them more work on the back end," he explains.

That means courts need to invest in their own plain-language forms, guided interviews, process maps, and FAQs, so that people find trustworthy, court-owned information before they go looking elsewhere. It also means being straightforward about what people need.

"Don't hide the ball," Zarnow says. "People are looking for information. Give it to them. Give it to them in a way that's useful... They don't need a PDF with every possibility — they need the PDF that applies to them."

Don't freak out

Judge Bailey, who has 31 years' experience on the bench, did not pretend the shift is small. "Don't freak out — just don't freak out," she notes. "I mean, look, I'm 67 years old. I'm a little freaked out. The possibilities of AI and the change it represents is a lot to take in.  It's a whole new world. But it's so cool. It is going to give us so many opportunities to help people." These tools, she adds, can reach people where they are, with less disruption to their lives, and give them a real shot at being heard.

That is the actual point of a justice system, accessibility for all, not screening people out based on polish. But instead, giving everyone who has a legitimate claim the chance to be heard and to receive a remedy.

As the webinar notes, those courts that measure their baselines now, standardize their business processes, and evaluate litigants on the merits of their case rather than the tools behind it will be better able to deliver on that promise at scale — and be much more prepared for what is coming.

For more on the impact of AI in courts, visit the TRI/NCSC AI Policy Consortium for Law & Courts‍