CommentaryPexels / Pixabay
Berkeley Law’s near-total default ban on student AI use gets the pedagogy exactly backward, a Maryland law professor and his student argue.
Berkeley Law recently announced a new artificial intelligence policy that, absent a professor’s decision to opt out, bans almost all AI use in students’ coursework and exams. Other law schools may soon follow suit. These bans teach the wrong lesson and should not become the standard policy at law schools.
Berkeley’s policy explicitly recognizes that “[f]uture lawyers may need to use artificial intelligence fluently.” It then proceeds to prohibit (absent opt-out by a professor) nearly every fluent use of AI. Students may not use AI to conceptualize, outline, draft, revise, translate or edit any work submitted for credit. They may not even ask an AI tool to correct a grammatical mistake. Instructors may opt out of the policy, and AI may be used in the narrow task of identifying sources for research, but the default rule is an almost complete ban.
We understand the impulse to ban AI, but we need to live in the real world, not the world we would like to live in. Regardless of what law schools decide, AI will be widely used by lawyers. Generative AI arrived in legal education faster than law school pedagogy could adapt. In law, we pride ourselves on the fact that we teach students to think like a lawyer. To us, that means we teach deep critical thinking and an ability to organize that thinking in a structured way. The prospect that students will outsource their thinking is terrifying to law professors and is a real concern as we seek to train the next generation of lawyers. The profession and our clients would suffer if lawyers fail to learn important critical reasoning skills. The cure, however, is training students to use AI thoughtfully and carefully. The failure to read cases suggested by AI, or to adopt AI’s reasoning without thought, is a mistake, but it is no different from mistakes many of us in older generations may have made as children by copying from an encyclopedia or, in the modern age, Wikipedia.
Our job is to teach the tool, not to pretend it doesn’t exist
The responsibility of law professors is not to bury our heads in the sand, but instead to reform the curriculum to teach students to be sophisticated users of AI. Recent graduates will enter a profession in which AI is already embedded. It lives in research platforms, in document review, in drafting tools their firms will require them to use, and in basic Google searches. Clients, too, will increasingly expect lawyers to be well versed in AI, just as they now expect them to use research tools like Westlaw and Lexis. Our duty as lawyers, captured in the ABA Model Rules, includes a duty of technological competence. The rule does not have an exception for lawyers whose schools discourage AI learning.
A law school that restricts supervised AI use for three years and then hands its graduates to a profession that demands it has not preserved rigor; it has exported the learning curve to the students’ employers. The first time our students critically evaluate an AI-generated legal analysis should be in a classroom, where the stakes are a grade and the feedback comes from a professor, not in practice, where the stakes are a client’s liberty or livelihood and the feedback comes from a sanctions order. Such orders are no longer hypothetical: courts have addressed AI-fabricated content in well over a thousand decisions, including a six-figure sanction this spring. And every one of those sanctioned lawyers was trained in exactly the AI-free curriculum that these restrictive AI policies now encourage.
Target the assignment, not the technology
There are genuinely good reasons to exclude AI from particular exercises or classes. A first-semester legal writing assignment might be designed so students struggle through their own first drafts, because the struggle is the point. A closed-book exam testing doctrinal mastery should be closed to AI for the same reason it is closed to treatises. But those are arguments for assignment-level rules, not an institution-wide default ban. Faculty routinely calibrate constraints to the skill being taught, and they can and should do so with regard to AI use.
Critical evaluation is the skill — and you cannot teach it through abstinence
The deepest flaw in these restrictive policies is that they misidentify where the cognitive work now lies. Increasingly, the lawyer’s task is not to produce a first draft from a blank page but to direct, interrogate, and evaluate machine-generated work product: Is this synthesis of the case law right? Did the tool miss the controlling authority? Is this argument actually sound, or merely fluent? That is legal judgment of a high order, and it is teachable, but only if students are allowed to practice it.
Most lawyers recognize this in practice. They routinely build on prior work. Lawyers rarely start from scratch. They use “brief banks,” form books, treatises, and other existing material. But one cannot simply copy a form book or file a brief-bank document unchanged; lawyers must research, revise, and tailor that work to a new situation. The same is true of AI’s output.
The pedagogical options are vast. Professors can provide students with AI-generated content and ask them to analyze and critique it. They can require students to submit the AI output they relied on, much as we might require an outline. And at times we should be demonstrating how to use AI as a productive tool: in a business law course, for instance, a professor might walk students through using AI to work through essential financial concepts like the time value of money, present value, future value, and basic valuation methodologies.
We recognize that Berkeley’s policy expressly invites opt-outs for those courses, and Berkeley has both included some teaching of AI in the legal writing class and has developed a particular course of study on the use of AI. But default rules matter and create both cultural and pedagogical norms. Relegating AI to specialty classes or to a few professors who are willing to engage with it treats the defining skill of the next decade like a niche course.
The grammar and spell-check prohibition is cruel
Finally, the Berkeley policy prohibits the use of AI to polish grammatical mistakes. For students with dyslexia and related learning differences, who now make up a meaningful share of every law school class, spelling and grammar tools are not a shortcut around critical thinking. They are the mechanism that lets thinking show. When spell-check arrived, some worried that students would never learn to use a dictionary; instead, a tool once considered optional became indispensable to every lawyer, judge, and law professor working today. Forcing a dyslexic student to submit work riddled with errors, or to spend hours with a dictionary, does not reveal the quality of her analysis or her critical thinking skills. Instead, it penalizes a weakness that will never matter in practice. We would never tell a student she may not wear glasses during an exam. Yet a rule that bars AI-assisted proofreading does precisely that to students. Yes, accommodations processes exist. But a policy that makes basic assistive technology a tool to be requested through a gatekeeping process rather than one presumptively available forces students to route a disability through the institution simply to use the digital equivalent of corrective lenses. We are training lawyers, not administering a grade-school spelling test. We should be seeking universal design in our classes, not creating requirements for students to seek accommodations.
The harder path is the right one: course-by-course rules tied to learning objectives, assignments that put AI output in front of students and demand they tear it apart, and an honest acknowledgment that the skill our graduates will be paid for is judgment about machine-assisted work, not the ability to pretend such work doesn’t exist.
Proponents of these restrictive policies remind us that critical thinking is the essential foundation of good lawyering. We agree. That is why we should be teaching students to think with and about these tools. It may be a heavy lift for professors as they rethink the curriculum and the way they teach, but that is our responsibility to our students and to the profession. Our students will practice law with AI. The only question is whether we taught them how.
In the spirit of the practice we advocate, we state clearly: AI tools assisted in drafting this essay. The ideas, the judgments, and the responsibility for every word are ours.
Donald B. Tobin is the Gordon, Wolf & Carney Professor of Law at the University of Maryland Francis King Carey School of Law, where he served as dean from 2014 to 2022. He specializes in tax law and election law and teaches Federal Income Taxation, Foundations in Business Law, Election Law, and Professional Responsibility. Before entering academia, he worked on Capitol Hill, clerked on the US Court of Appeals for the Fourth Circuit, and served as an appellate attorney in the Tax Division of the US Justice Department.
Samuel Irwin is a JD student at Maryland Carey Law and Professor Tobin’s research assistant.
Opinions expressed in JURIST Commentary are the sole responsibility of the author and do not necessarily reflect the views of JURIST’s editors, staff, donors or the University of Pittsburgh.
