Against that backdrop, a spectrum has emerged between unlimited AI use and total bans, with two early frontrunners — the University of California, Berkeley School of Law at the more restrictive end, and the University of Chicago Law School in the middle — illustrating two possible paths forward.
The schools are setting the pace at a time when educational institutions all over are increasingly looking for guidance on how to proceed, experts said.
"I fully expect within five years the majority of American law schools to adopt some policy like one of these two schools," said Ed Walters, vice president for legal innovation and strategy at legal AI company Clio and an adjunct professor at the Georgetown University Law Center and lecturer at the University of Chicago Law School. He also sat on a committee that helped design UChicago Law School's policy.
"Law schools will have to choose how to deal with generative AI, and maybe they take one approach or the other, maybe they modify one or the other, but everyone's going to have to have some policy," Walters said. "These are really just two of the first most comprehensive ones."
Starting next academic year, UChicago Law School will curtail AI use among students, especially first years, including through an attention-grabbing move to block devices from classrooms, it announced in July. But it will give students and instructors more room to experiment in upper level elective courses.
The school got feedback from faculty, alumni, law firms and others about the need to teach students independent and rigorous legal thinking, but it also took into account the need to teach students to engage with AI ethically, Dean Adam Chilton said in an interview with Law360 Pulse.
"We're trying to accomplish both of those goals," he said.
UC Berkeley Law School, by contrast, made a splash in May with a more restrictive policy that bars virtually all uses of AI on assignments submitted for credit, including "conceptualizing, outlining, drafting, revising, translating or editing," as a default rule, though professors can adopt their own policies.
Dean Erwin Chemerinsky told Law360 Pulse in an interview that UC Berkeley Law School still hopes to train future generations of lawyers that are fluent in AI skills needed in modern practice.
"But when a student is turning in a paper or an exam for a grade, it should be the student's work that's being turned in," he said. "A student shouldn't be turning in for a grade an exam or a paper that's written by another student, or by a lawyer, or by [Anthropic's AI tool] Claude."
Differing Approaches
UChicago Law School's toughest restrictions on AI use come during the 1L year, when law school curricula across schools have historically focused on the fundamentals of legal argument through the Socratic method of class discussion, Chilton said.
Students will be prohibited from using devices in classrooms, with some exceptions, and exams will be conducted without access to the internet. Chilton said other schools at UChicago, including the business school and public policy schools, already had policies like the device ban in place, and there was no practical way to allow devices in class while preventing them from being used for AI.
Students will be allowed to use AI for research but not writing in their first-year legal writing work, and the law school added an oral requirement for a major writing assignment for second-year students.
As students get into upper-level electives, instructors will have more latitude to incorporate AI if they want, and students will be exposed to AI through the law school's legal clinics.
Chilton said the overarching purpose is to get away from testing early law students on their rote skills and to focus on the arguing and communication skills that are the backbone of lawyering.
"It's not, 'Can you memorize tort doctrine?' It's, 'Can you get good at that skill of argumentation?'" Chilton said. "That's something that I think it really matters to try to do — to think about how we can ensure that we're giving the students the best opportunity for learning to think like a lawyer in that critical way."
At UC Berkeley Law School, it is taking a harder line: banning AI as a default rule from uses like brainstorming or outlining a paper, summarizing a legal rule for use on a class assignment, or even revising a paper to fix grammatical errors. The school adopted its policy in May, and it is already in effect for this summer.
The default is just that: a default, Chemerinsky emphasized. Instructors have "complete" authority to customize AI rules for their particular courses, although it's too soon to see for sure how they'll exercise that authority, he said.
"Based on practice, I think there's going to be huge variation among faculty," he said. "Some are very much wanting to incorporate AI into their class and into how students produce material. Some feel strongly that they don't want the students using AI."
Chemerinsky rejected criticism that students could make it through all of UC Berkeley Law School without encountering AI. AI has been incorporated into the first-year legal writing class, and the school offers courses on AI-related issues, he said.
At the same time, Chemerinsky said it was important for the school to make sure that law firms and other employers can have confidence that the assessments UC Berkeley Law School students receive are based on their own abilities, not on AI.
"When employers look at a student's grades, and a lot of lawyers do, they want to make sure that the grades are reflecting the students' work," he said.
Outside Perspectives
Both law schools have quickly received pushback over different components of their policies — for UChicago Law School, often centered around its device ban, and for UC Berkeley Law School about the scope of its policy — as well as support in the legal academic world.
Daniel Linna, a law professor and director of law and technology initiatives at the Northwestern University Pritzker School of Law, doubted whether instructors at UC Berkeley Law School would fully take advantage of their options for customizing AI policies beyond the school's default no-AI stance.
"We know that these defaults are sticky," he said. "If you say, 'Oh, the default is students can't use AI,' that relieves faculty from the burden of thinking about, 'Well, how should AI be used responsibly in my course?'"
Donald Tobin, a faculty member and former dean of the University of Maryland's Francis King Carey School of Law, called UC Berkeley Law School's AI prohibition "Draconian." He said he was more comfortable with UChicago Law School's approach, but questioned the ban on using devices in class.
"It strikes me that we are teaching graduate-level adults who are about to be into a profession where they are charged with upholding the Constitution, stopping people from going to jail, defending people in court, and the idea that we can't let them use a computer in class, to me, it's a mistake," Tobin said.
Howard Wasserman, a professor and associate dean for research and development at the Florida International University College of Law, said another way to think of the divide between the two schools is that UChicago Law School is taking an institution-wide approach, while UC Berkeley Law School is giving more priority to individual instructors' priorities.
Wasserman called AI a "genuine threat" to the practice and education of law, and he said he bans it from his classes. He said he expects to see more schools follow UChicago Law School's approach, although he personally favors UC Berkeley Law School's.
"I would love it if we could all just do what Berkeley's doing," Wasserman said, "but I don't know that that's possible."
Lakshita Bhargava, a master of laws student at UC Berkeley Law School, told Law360 Pulse that from the student perspective, there is a wide range of ability to use AI. Students are concerned that more-skilled students will be able to get around the ban, and that AI detectors are not reliable enough to consistently catch its use.
A more comprehensive policy than UC Berkeley's would "let students distinguish between passive and active uses, require meaningful disclosure, protect confidentiality, mandate verification, and preserve full student responsibility," Bhargava said.
Other experts disputed whether UChicago Law School's and UC Berkeley Law School's approaches were as different as they appeared at their cores.
Walters said that looking closely at the schools' written visions — with UChicago Law School's focus on "AI-resilient pedagogy and assessment" and UC Berkeley Law School's statement that "thinking remains the sine qua non of good lawyering" — shows they have more in common than they might appear to.
"The goals are very much the same," Walters said. "UChicago's take is a little bit different, but they are both making a difficult stand at a very fast moving time."
Walters also said that from his perspective in the legal tech industry, the companies behind legal AI are generally supportive of law schools forming some kind of plan for student AI use, even if it limits that use in some ways.
"We are rooting for law schools to teach the responsible use of AI tools," he said. "The more they teach, the better the adoption of these tools is going to be, and I think that the unethical and irresponsible use of AI hurts technology companies more than anything at all."
The question of how to handle student AI has prompted schools to think about the fundamental purpose of legal education. Chilton said the planning process behind UChicago Law School's policy raised difficult questions.
"What is the thing that we should be teaching?" he said. "What's the thing that's essentially human? What's the value of our education? And how can we ensure that we're preparing our students for the marketplace they're going to enter?"
Chemerinsky described a similar grappling at UC Berkeley Law School, and a policymaking process that is on the horizon for all law schools.
"I think every law school, and I think every educational institution, is going to face the same tension," he said.
"On the one hand, we need to train our students how to effectively responsibly use AI. It's an incredibly powerful tool," Chemerinsky said. "But on the other hand, we might make sure that when students are turning in work for a grade, it's their work and not produced by AI."
The two schools may have staked out different territory on the spectrum of how to approach AI, but according to Walters, neither option is as fraught as failing to plan for AI — or in other words, "training law students as if it's 1987."
"The most controversial stance of all," Walters said, "is to do nothing."
--Editing by Adam LoBelia.
For a reprint of this article, please contact reprints@law360.com.
