University of Texas Law Alters AI Policy to Stop Skills Loss (3)

The University of Texas School of Law’s dean is calling for a new approach with generative artificial intelligence policies to counter the risk that the technology will erode the learning of essential skills.

Faculty should emphasize Socratic-style questioning in teaching, Dean Robert “Bobby” Chesney said in a Tuesday memo to staff. They should view classroom time as “the sole context in which a professor can be certain” a student is learning without AI.

Professors should “emphasize to students the value of doing the hard work in the first instance themselves,” Chesney said, though also ensure they learn AI skills. “The lawyers most likely to flourish in the years ahead are those who both are adept at making wise use of AI capabilities and possess the same expert judgment and perspective that have always been the hallmarks of the best lawyers,” he said.

The memo comes as law schools across the US are grappling with how to teach students about ethics rules and legal writing as the future lawyers gain access to rapidly evolving AI tools.

The prestigious Texas law school in the memo fashions its response to AI in three distinct areas: deciding what AI skills to teach students, protecting the integrity of graded assessments, and guarding against students using AI as a substitute for the work of reading and reasoning through legal materials themselves.

The document’s sharpest argument concerns “cognitive de-skilling"—the notion that a student’s ability to think might be eroded by the use of tech. AI is “considerably more tempting” than past shortcuts like commercial study aids, given its “comparative availability, confidentiality, and range, as well as its distinct capacity for iterative dialogue,” Chesney said in the memo.

“In the emerging age of AI, our students will flourish best if they possess the fruits of both traditional legal training and a thoughtful program of AI upskilling that will help them make the best possible use of AI once in practice,” he said.

In response to an e-mailed question, Chesney said that “our approach is not top-down in the sense of mandating one-size-fits-all.” He added, “The idea is to make sure our colleagues are well-informed and actively thinking about the complexities, and sharing ideas with each other about creative solutions.”

AI Challenges

Austin-based UT Law, which tied for 16th place nationally in the latest U.S. News and World Report rankings and has more than 900 enrolled students, has already imposed a “near-complete elimination of take-home exams” in favor of in-class tests, requiring the use of “software limiting what can be done,” Chesney said.

The more difficult challenge is the writing seminar, Chesney said. “With ubiquitous access to AI, the temptation to get external help is significantly greater,” he said. Some faculty have responded by asking students to document their AI use in their writing process rather than banning it.

At UT Law, Chesney said he rejects the idea that there’s “irreconcilable tension” between educational goals and the pressing need to train students to use AI. “These goals are not in a zero-sum relationship,” he said in the memo.

The law school became one of the first to give students licenses to AI platforms including Harvey and Legora, and it expects more licenses in the near future, Chesney said. The school also plans that all students, faculty, and staff will get “enterprise-grade licenses with leading frontier labs this fall,” with AI learning objectives mapped onto required courses including 1L legal research and writing and upper level professional responsibility courses.

Law schools have been taking differing approaches to AI. UC Berkeley School of Law’s AI policy, which goes into effect this summer, forbids students from using AI to conceptualize, outline, draft, revise, edit, or translate work submitted for credit.

Columbia Law School’s policy, in effect since August 2025, prohibits generative AI entirely on exams or graded papers, including “for aid in drafting any part of work submitted for credit, even if the use is fully documented.”

The University of Chicago School of Law’s policy bans AI on exams but allows the tech for brainstorming or proofreading for papers and other graded work. The university is rolling out mandatory AI modules for all first-year students this year “to bring everyone up to a minimum level of literacy with generative AI.”

AI Fixes

Daniel W. Linna Jr., a professor and director of law and technology initiatives at Northwestern Pritzker School of Law, said that de-skilling is “a well-documented problem,” but he argued the fix is building better AI tools.

Socratic dialogue is valuable but isn’t scalable: a student might get called on only “one, two, or maybe three times,” during a course, he said. Law schools and professors should develop AI tools to “tutor and mentor students,” offering them “unlimited repetitions,” Linna said.

Julian Nyarko, a professor and co-chair of Stanford Law School’s AI Initiative, said evidence on de-skilling is “complicated,” with randomized controlled trials producing conflicting results. A Cornell University study finding students using AI on a first task “leads them to perform worse and give up sooner on subsequent tasks,” he said. However, a similar studyat the University of Minnesota Law School, he said, found “no detrimental effects of AI access on subsequent, unassisted tasks.”

“There probably is some risk of cognitive de-skilling if AI is implemented carelessly, but we shouldn’t forget that there is also tremendous opportunity if done right,” Nyarko said. Recent AI models can answer questions “better than most law professors can on the spot” and offer scalable learning experiences like simulation exercises that have “previously been impossible.”

Still, Nyarko said no school has found a way to push cheating risk to “near zero” short of measures used by companies like keystroke logging or gaze tracking, which he said schools have shown little interest in adopting.