The New Pro Se: Generative AI and the Surge in Federal Civil Self-Representation
This paper analyzes federal civil litigation data from 2008 to 2025 to demonstrate that the widespread adoption of generative AI has driven a significant increase in pro se filings, particularly among first-time and female plaintiffs in civil rights cases, resulting in more citation-dense complaints that face higher dismissal rates without improving litigation outcomes.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
The Big Picture: A New Tool for the Courtroom
Imagine the federal court system as a massive, complex library where people go to file official complaints against others. For a long time, if you didn't have a lawyer (a "pro se" litigant), writing a complaint was like trying to write a novel without knowing grammar or plot structure. It was hard, expensive, and many people gave up.
Then, in late 2022, a new tool arrived: Generative AI (like ChatGPT). This tool acts like a super-smart, instant ghostwriter. The paper asks: Did this new tool change who is walking into the library, what their complaints look like, and whether they actually win?
The short answer is yes, but with a twist. The tool helped more people get their foot in the door, but it didn't necessarily help them win the game.
1. The Surge: More People Filing
The Finding: Before AI, about 11% of federal civil cases were filed by people without lawyers. After AI became popular, that number jumped to nearly 17%.
The Analogy: Think of a popular concert venue. Before the new ticket app (AI) came out, only 11% of the crowd managed to get in because the line was too long and the process too confusing. Once the app launched, the line moved faster, and suddenly 17% of the crowd got in.
The paper shows this wasn't just a one-time bump; the number of people filing without lawyers kept climbing steadily, especially in cases involving Civil Rights and Other Statutory laws. It wasn't just a few repeat offenders using the tool; it was a wave of new people entering the system.
2. The "AI-Flagged" Complaints: Polished but Hollow
The researchers couldn't ask people, "Did you use AI?" so they looked at the text of the complaints to find "fingerprints." They looked for specific patterns, like:
- Sentences that are all the same length (too perfect).
- A sudden explosion of legal citations (referencing laws) that look impressive but might be generic.
- Specific vocabulary that AI loves to use.
The Finding: About 14% of the new, non-form complaints (those not using pre-printed court forms) showed these "AI fingerprints."
The Analogy: Imagine two students submitting essays.
- Student A (No AI): Writes a rough, messy essay with some grammar errors but tells a very personal, specific story about their own life.
- Student B (AI): Writes a perfectly formatted essay with fancy vocabulary and a huge list of references. It looks amazing on the surface.
The paper found that the "AI students" (AI-flagged complaints) looked much more polished and citation-heavy than the "no-AI students."
3. The Paradox: Better Looking, Worse Results
Here is the surprising part. You might think that if a complaint looks more professional and has more legal references, the judge would be more likely to listen and rule in your favor.
The Finding: The opposite happened.
- Dismissals: The AI-flagged complaints were more likely to be thrown out by judges (61% vs. 53%).
- Early Exit: They were more likely to be stopped at the very first hurdle, before the other side even had to answer.
- Win Rates: There was no improvement in winning the case. In fact, they won slightly less often, though the difference wasn't statistically huge.
The Analogy: It's like a beautifully wrapped gift box that is empty inside.
The AI helped people wrap the complaint in shiny, professional-looking paper (citations, perfect grammar). But the "gift" inside—the actual legal argument and the specific facts of the person's story—wasn't any stronger. In fact, because the tool made it so easy to file, people might be filing cases that are legally weak or don't belong in court at all. The judge sees the shiny wrapper, opens it, and realizes the contents don't meet the legal requirements.
4. Who Is Using This Tool?
The paper looked at who was using this new drafting power.
- First-Timers: The AI users were mostly newcomers. They weren't the same people who had been filing lawsuits for years. It seems the tool lowered the barrier enough for people who had never tried to sue before to finally take the leap.
- Gender: There was a small, suggestive hint that more women might be using the tool compared to before, but the data isn't strong enough to say for sure.
- Money: Surprisingly, the tool didn't seem to be used primarily by the poorest people (those who ask to waive filing fees). The surge in filings happened across the board, not just among those who couldn't afford court costs.
The Analogy: Imagine a gym. Before, only the serious athletes (repeat filers) or the very poor (fee-waiver users) showed up. Now, with a new AI personal trainer app, a bunch of regular people who had never worked out before are showing up. They look great in their new gym clothes (polished complaints), but they aren't necessarily lifting heavier weights (winning cases) yet.
5. The "Asymmetric Adaptation" Problem
The paper ends with a warning about the courts.
The Finding: People can download and start using AI tools instantly. Courts, however, are slow. They have rules, budgets, and ethics committees that take years to approve new technologies.
The Analogy: Imagine a highway where cars (litigants) suddenly get turbo-charged engines (AI). They are zooming onto the road at 200 mph. But the traffic cops and toll booths (the courts) are still using old, slow equipment to check tickets and manage traffic.
The result is a traffic jam at the entrance. The courts are getting overwhelmed with a flood of new, shiny-looking complaints that they have to spend time reviewing and rejecting. The system is getting clogged because the drivers are moving faster than the road can handle.
Summary
- What happened? Generative AI made it much easier for regular people to write and file federal lawsuits.
- The Result: More people are getting into court (Access to Drafting).
- The Catch: These new complaints look better on paper but are actually getting rejected more often. The tool helped people enter the door, but it didn't help them win the room.
- The Lesson: Technology can change how institutions work from the bottom up, even if the institutions themselves haven't adopted the technology yet. Courts need to adapt to handle this new flood of "polished but empty" complaints.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.