Over the course of his career, Joseph McMullen has dealt with some of the most powerful agencies in the country: the FBI, Customs and Border Protection, and Immigration and Customs Enforcement. But in early 2024 the San Diego–based civil rights attorney faced a problem of scale. He had three federal trials in three months—two involving deaths in jail, one involving American children detained at the border—and terabytes of documents. He turned to artificial intelligence to help him get through it all.
McMullen’s path to the courtroom has been unconventional. A former analyst at the consulting firm Bain & Company, he received a law degree at the University of Virginia and trained at the Trial Lawyers College (now called the Gerry Spence Method) in Wyoming in a program that specialized in the emotional craft of storytelling. The emphasis he places on both analytical rigor and narrative instinct has led him, unexpectedly, to artificial intelligence.
Scientific American spoke to McMullen about how AI can free lawyers to focus on what makes us human.
On supporting science journalism
If you’re enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
[An edited transcript of the interview follows.]
You litigate civil rights cases against local and federal law enforcement. What does that look like on the ground?
They often involve violence—shootings or tasings. One client was diabetic and had a seizure at a restaurant. Police arrived and jumped to the conclusion it might be drug-related. They tased and arrested him, notwithstanding paramedics saying his blood sugar had crashed.
Other cases involve deaths in jail—failures to follow rules about keeping inmates safe, whether inmate-on-inmate violence, physical abuse by employees or failure to provide needed medical care when there’s clear signs someone’s in distress.
When did you turn to AI for help?
Early 2024—I had looked at ChatGPT before to see if it could find me a case, and it hallucinated the perfect case, but it wasn’t real. ChatGPT came clean and said the case with the full citation didn’t exist. That ended any interest that I had in using AI, probably in late 2023.
But with these impending trials, I thought, there are tasks I’m performing that aren’t the best use of my time. So I started exploring AI platforms like Clearbrief and Briefpoint.
What I’ve found is that to put together a successful trial, there are a few things you have to do. First, gather all the stuff your case might be about—documents, location data, photographs. Second, figure out what your case is about. A lot of that analysis can be done by AI. You can feed it data and have it break it down.
But lawyering is also about judgment—that part can’t be farmed out to AI yet. So third, tell the story of your case in a compelling way. That’s the human element. By getting help with gathering and analysis, it frees up time to focus on discovering the story AI can’t find. It can analyze 100,000 text messages and give me an understanding of what’s relevant so I don’t have to.
Can you give an example where AI changed the outcome of a case?
We represented a girl named Julia and her brother Oscar, both U.S. citizens. [In 2019, when Julia was nine years old and Oscar was 14], Julia was accused of being an impostor, an undocumented cousin [whom Oscar was accused of attempting to smuggle] across the border. Oscar was held for 14 hours, and Julia [was held] for 34 hours, much of [that time being spent] in an underground jail. Eventually Telemundo caught news [of this]. When reporters showed up, folks at the border realized what was happening and let them go.
That was a five-year battle involving a lot of documents. I used Clearbrief to put together a brief hyperlinked to evidence that AI helped me sort through. The judge gave our clients a substantial verdict, with discussion about how what happened was incompatible with our values.
Have you used AI for strategy rather than just sorting evidence?
Yes. Another example: a successful jail death trial in May 2024. That case recently was upheld on appeal. To help my co-counsel prepare for oral argument, I uploaded the appellate briefs and record into the software CoCounsel and asked it to put together an opinion in which we lose on every issue—the best analysis of why we lose. It generated the opinion. I circulated it so we could be prepared for the best arguments against us. Just because you out-lawyer the other side doesn’t mean you’ll win—judges have their own research staff and wisdom. My co-counsel did a phenomenal job and very much appreciated having that opposing argument. All that work was done in less than a minute.
What is your philosophy on how lawyers should use this technology?
First, verify everything. If AI cites a case, read it. And never upload confidential information without guarantees that it won’t be used for AI learning.
Beyond that, think about what advocacy is. Aristotle told us effective advocacy needs logos, ethos and pathos—logic, credibility and emotion. Every aspect of lawyering involving logic, AI can help us [with]. The credibility part—being thorough, reviewing everything—AI can help with that, too. But emotion—I don’t think AI can add that. Emotion is finding real human connection with issues that resonate with all of us. Each of these cases has been about love, betrayal, loss and joy.
Use AI to help with any logical task. Farm out the logical analysis and gathering. You’ll free up time to understand the emotional story only a human can. That’s what makes AI great: not helping lawyers turn into robots but helping lawyers focus more on the humanity of what we’re doing.
A version of this article appeared in the March 2026 issue of Scientific American as “Joseph McMullen.”