OK, look, we know you haven't signed up for our brand new indicator newsletter yet, but it's fine.
It's fine.
It's not a big deal.
Vito, it sounds like it is a big deal to you.
Just look at the show notes to sign up.
All right.
It comes out every Friday morning.
N.P.R.
Like many attorneys, Laura McClain is a bit of a Luddite.
I keep paper files.
I'm kind of old fashioned, even though I'm not that old, right?
And Laura was at this trial earlier this year.
She does family law, you know, divorces, custody disputes, heavy stuff.
On the opposing side was a woman without an attorney representing herself.
So going pro se, as it's called in the biz.
Yeah.
And it was a divorce case involving finance records and thousands of dollars at issue.
And it's going really badly.
The woman on the other side is forgetting to object to things.
She doesn't have the right number of copies of documents for the court.
She's getting flustered and things are just generally falling apart.
By the second day of the trial, she can hardly speak.
It was an absolute disaster.
She wasn't even able to present evidence just because she had relied so heavily on this outline.
At one point Laura offered to help her find her documents from her outline and give it to the judge, just to move the trial along.
She'd been rifling through boxes and boxes of loose documents.
And then, as Laura is helping, she catches a glimpse of the woman's outline, her blueprint for winning the case.
It's all AI-generated.
Even the chatbot prompts were still in there.
And in the end, she lost the case.
This is The Indicator from Planet Money.
I'm Vito Emanuel.
And I'm Darian Woods.
AI is transforming the legal system.
And today we're going to talk about the wider costs of making it much easier to file a case by yourself without a lawyer.
On the one hand, people who can't afford lawyers have a legal coach in their corner.
On the other hand, it's not a very good coach and it's clogging up the courts.
Joshua Levy used to think AI wouldn't have that much of an effect on the legal system, but his friend Anand Shah thought no way.
Clearly, AI was going to touch everything.
Joshua and Anand are PhD students at the University of Southern California and MIT, respectively.
So they decided to settle their debate with research.
Putting numbers to their reckons.
I love it.
And they pulled down a bunch of data from the federal courts on pro se court filings as one way to measure this.
And we just saw these really dramatic results in filing numbers and our jaws just kind of collectively dropped.
That's Anand relishing in his sweet victory.
Pro se cases are up, way up.
Starting in 2023, right around the release of ChatGPT, the share of pro se cases jumped.
Now, this is just for federal courts where the data is easy to analyze.
We don't know about state and local courts, but it might be even more pronounced there.
And there's this thing that's happening with the quality of lawsuits being filed.
Since pretty much forever, it has been hard to file a lawsuit.
Just knowing what to do, what paperwork to file, what claims to bring, and so on.
So only certain people followed through.
The people that were willing to climb all the hurdles before felt in some sense that you know their cases were especially strong, or something like that.
People file pro se cases for a bunch of reasons.
Sometimes they can't afford an attorney.
Sometimes something has gone really wrong in their lives.
An attorney won't represent them and they just want to have their day in court.
AI is lowering the barriers for all those people.
But certain types of cases are especially AI-friendly.
Anand and Joshua call them templatable.
These are things like foreclosure cases or housing cases.
At local levels.
It would also be disputes that wind up in small claims courts cases with relatively straightforward legal questions.
Things a court might literally have a template for on its website.
And our PhD students said for templatable types of cases, they've seen not only a rise in the number of cases filed, but the number of documents filed in each case.
Which, of course, means more work for the courts.
Backlogs in the courts have been a problem for years now, even before ChatGPT.
Judges still only have 24 hours in a day.
You need courtrooms, you need clerks, you need all these things to make a court really work.
The supply of court services is not really changing.
Not that people who go pro se are winning any more often with AI advice.
Here's Joshua.
It's still extraordinarily hard for a pro se litigant to bring a case against the government or against a private company.
The share of unambiguous wins for pro se litigants is still essentially zero.
Remember Laura's story about the pro se litigant?
Filing a lawsuit is just step one.
It's a big step, but the steps really just get bigger from there.
Like presenting evidence, making the case with your voice out loud in the courtroom.
Historically, judges try really hard to be charitable to pro se litigants because the odds are so long.
But recently some of the litigants have been getting hit with thousands of dollars of penalties, thanks to AI.
Yeah, one pro se was ordered to pay over 60000 in fees, in part because he cited a bunch of bogus AI-generated cases.
Right, so pro se AI lawsuits are starting to clog up court dockets.
There are more cases filed, and the people filing them aren't doing much better than pro se litigants before AI.
Some are getting fined, so those are some of the downsides.
But there is this... other argument, the case for AI help.
By this view.
AI is helping the people who are left out of the legal system have their day in court addressing an access issue.
Right.
So let's meet Satish Nori.
This is an opportunity for the courts to say, we're going to do better.
We're going to address these issues.
Satish was an attorney at a legal aid organization in New York for many years and now researches AI applications in the law at NYU.
And he remembers all the people he couldn't help at his old organization.
There just weren't enough attorneys.
I saw hundreds and hundreds of people every year who are facing addiction and yet we're turning away half the people who are facing addiction.
So it hit me that like, this is the fee and we're never going to meaningfully bridge this justice gap.
AI might be an imperfect way to address this, but he says it's probably the best shot right now at doing so.
Satish points to how AI can help out with all the legal-y things before filing a lawsuit.
That attorneys could help answer, but don't have enough time in the day to right now.
Like how to write a demand letter to your landlord.
Or if your heat goes out in the winter and you don't know what to do, a chatbot could tell you to Call the city.
Contact your landlord in writing.
Maybe you should contact your neighbors and organize your building.
Maybe help them contact a tenant organizer.
But when it comes to actually filing a lawsuit, Satish says open the floodgates.
Anyone and everyone should be able to file a lawsuit in the easiest way possible.
There's so much unmet need, right?
So we should let everybody file.
He's like the legal version of Mel Robbins.
Let them, parentheses, sue.
Yeah, self-help for the AI legal age.
And if AI gets stuff wrong or makes stuff up?
Satish says judges should take it even easier on people going pro se who wouldn't otherwise be able to file a lawsuit without AI help.
And as for courts grappling with new case volume as the result of that, Satish says well, that's like their job.
This is a kind of irresistible force that's being met with an immovable object.
The courts right.
And the force is all the unmet needs of Americans who have legal claims and it runs smack up against the courts.
Still, right now, the courts don't seem like they have the resources to handle a flood of cases, nor do they seem like they're willing to lax AI-related penalties.
And a bigger backlog means people who need immediate relief like say, who are asking for a protective order against an abusive partner, might have to wait longer for that relief.
Right.
This feels like there is an excess and deficiency tradeoff right now.
Yeah.
And Laura McLean, our family law attorney from earlier.
She's watching this play out in real time right now.
One client of hers kept second guessing her advice in a custody case with an AI chatbot.
Eventually, he decided to go pro se and use AI instead of Laura.
He lost custody.
Yes, he lost parenting time as a result because he wasn't willing to take advice.
In some cases, Laura says, you should probably just stick with a real flesh and blood attorney.
This episode was produced by Cooper Katz McKim and engineered by Sina Lafredo.
It was fact-checked by Sierra Juarez.
Alex Goldmark edited this episode, and Cade Kincannon is our show's editor.
The Indicator is a production of NPR.
