Legal aid and AI help poor Americans close ‘justice gap’

Washington — From pen and paper to artificial intelligence (AI) and algorithms — the path to legal redress is growing faster for many poor Americans as AI helps close a yawning US justice gap.

Be it fighting evictions or domestic violence, the sort of bureaucratic logjams that often stymie low-income defendants could ease as AI bolsters overstretched legal aid teams once armed with little more than a lined, yellow pad.

“Generative AI has the potential to do more to improve access to justice than anything I’ve ever seen,” said Jim Sandman, a lecturer at the University of Pennsylvania Carey Law School.

Take the low-slung legal aid office that sits east of the coast in San Bernardino, California, an outfit that has seen its workload transformed by tech.

Before AI, centre director Pablo Ramirez said there was no way for his team to keep up with the myriad needs of the 200,000 low-income Californians they serve, his offices jammed with eager volunteers, dated desktops and reams of paperwork.

He said legal cases — of whatever kind — often meant hours of in-person interviews and endless form-filling.

But Ramirez thinks help may now be arriving in the form of new AI tools that automate some of the more cumbersome processes — and he is working all out to get his 44-person office ready.

“Before, this was a pen-and-paper organisation,” said Ramirez, executive director at the Legal Aid Society of San Bernardino, a racially diverse community east of Los Angeles and the largest US county.

“People would line up outside our door, wrapping around the building ... to meet with an attorney and get their documents prepared.”

The situation only worsened during the pandemic.

“We have had to figure out how to continue to serve this growing population in our counties with limited resources. So I had to bring in technology,” he said.

Justice gap

The legal needs of the poor far outstrip the help on offer, a “justice gap” affecting 92% of legal problems experienced by low-income Americans, according to the Legal Services Corporation, a major provider of legal aid services.

Among the world’s 46 richest countries, the US ranks last on both access to and the affordability of its civil justice, according to the World Justice Project.

Available resources from the government and grantees “are grossly inadequate to meet the demand for legal services,” said Ronald Flagg, president of the Legal Services Corporation.

Yet the efficiencies that could be gained with new AI tools, he said, could be a “game changer”.

For Ramirez, initial AI experiments are already helping his team with accounting, case management, document preparation — even finding new pro bono volunteers.

“It can take an interview that would (take) an hour or two hours to do, plus documents, and knock it down to 15 minutes,” he said.

Whereas in previous years the office served about 2,500 people a year, that has now risen to 8,000-10,000, he said, all the while lowering burnout rates among his colleagues.

Then there’s cost

Alongside AI’s many positives, there are clear risks to its use in sensitive settings, said Pennsylvania lecturer Sandman, pointing to privacy considerations and simple correctness.

But Sandman said he was less concerned by such issues, given that high-powered law firms had just as much incentive to get these considerations right.

More of a question, Sandman said, was how to get these tools into the hands of legal aid offices — and not just private law offices — a failure that could otherwise widen the justice gap.

“The best generative AI tools are not cheap,” said Sandman, who also co-leads a task force on AI and the legal profession for the American Bar Association.

“The risk is that generative AI might end up increasing the gap between the resources that are available for clients of means, and the resources that are available to those who don't.”

Big firms could widen access to their tools, similar to the pro bono model for legal aid, he said, or technology companies could make their tools available free or at a reduced price.

In a report in August, the task force found that generative AI in particular holds promise about access to justice, but warned that price poses a major hurdle.

Putting non-profits on a level playing field is a primary goal for Everlaw for Good, created in 2017 by the legal software company Everlaw and which recently announced expansion plans.

“Generative AI is a sea change in the way people think about AI. It’s quite transformative and will change how legal teams work,” said Everlaw founder AJ Shankar.

Everlaw for Good tools, made available to non-profits, journalists and others, including the Legal Aid Society of San Bernardino, can help advocates take on more clients and pursue major litigation, said Joanne Sprague, head of Everlaw for Good.

“These cases are five to 10 years long, and the other side has unlimited resources,” she said.

“The ability to even commit to class actions or mass torts, where there can be hundreds or thousands of clients — without technology, they would have to be turned away.”

Putting it into action

For the most part, the public is not yet turning to popular AI platforms with legal questions, said Margaret Darin Hagan, executive director at Stanford University’s Legal Design Lab.

The lab has been talking with legal aid groups and locals to gauge their openness to such tools.

So when the team asked people to try out AI in a fictional reviction scenario, “people saw a lot of helpfulness, even if they knew they couldn't trust it or were wary,” she said.

Users said AI helped them figure out options, rights and who could help.

Users had typically searched online for that sort of information, which Hagan said tended to yield useful general fodder but gave less help on local resources — a gap she said AI tools could hopefully help fill.

Help with next steps has been a key focus for the JusticeBot that Hannes Westermann created with housing authorities in Quebec, aiming to help tenants in housing disputes.

It has been used tens of thousands of times since 2021.

“There’s a big gap in people encountering a situation and realising it’s a legal situation — they don’t know where to start,” said Westermann, who is now an assistant professor in AI and law at Maastricht University in the Netherlands.

The tool he built “doesn’t tell them, ’You should do this,’ but it lets them know they have rights in this situation.”

He is now working to incorporate generative AI, hoping it can play a role in legal mediation, along with building tools to help legal aid groups deal with excess paperwork.

Filling out legal forms is a sweet spot for AI applications, Westermann said, as it involves large volumes of work but only basic legal reasoning.

“Lots of people need to deal with this very frequently, and at the same time often people are underrepresented,” he said.

“At the moment, the alternative for a lot of people is just having nothing, so integrating AI could be very beneficial.”

Thomson Reuters Foundation