In many university computer science programs, students are not only allowed to use artificial intelligence during exams, but they are also encouraged to learn about the benefits and efficiency that AI provides to the coding process. However, during exams, the students must also explain how and why they relied on AI, walk through their coding decisions, and demonstrate they understand the functionality.

Unfortunately, law schools take varying approaches to using AI and may not equip students for the reality of legal work they will face. In fact, just last month, the University of California at Berkeley School of Law instituted a restrictive AI use policy with prohibitions that “include (but are not limited to)” using AI to brainstorm, suggest the structure of a paper or translate documents into English.

The reality is lawyers use AI to conduct research, draft documents, review contracts and handle large volumes of information. As students graduate and become lawyers, they will encounter these tools immediately. Without training, they will not be prepared to use them with sound judgment, creating a readiness gap among the next generation of legal professionals.

I recently spoke with professors at Indiana University’s Luddy School of Informatics, Computing and Engineering who specialize in artificial intelligence. In their courses, AI is part of the workflow. Students use it in their development process, and grades depend on whether the student can clearly explain their reasoning and understanding of the output. This approach is not isolated to Indiana University, and computer science programs are increasingly incorporating AI into both coursework and assessment.

By contrast, at law schools, policies on using generative AI tools vary widely. Although headway is being made, many schools do not meaningfully build AI into how students are taught or assessed.

At the same time, leaders across the legal industry are assessing the evolution of tech in legal education.

For example, Austen Parrish, the University of California at Irvine School of Law dean and chancellor’s professor of law, recently predicted that legal education will continue to integrate generative AI. The chief of staff to the general counsel and head of legal operations at Organon, Stacy Lettie, put it more directly: “The work has evolved, so learning must evolve with it.”

Much of the concern around AI in law school comes from real examples. Courts have repeatedly sanctioned attorneys for submitting filings that included hallucinated case citations. Even federal judges have admitted to issuing faulty rulings stemming from generative AI use by their staff. Critically, the issue at the core of these incidents is that the output was never verified.

There is a cognitive bias known as automation bias, which describes the tendency to trust suggestions from automated systems. Indeed, in a recent study, theoretical neuroscientist Vivienne Ming asked humans to make predictions about real-world events using scenarios drawn from a prediction market platform. Participants who were permitted to use AI generally used the AI answer as their own. Importantly, those who fared best in the study were what she called “validators,” who questioned and pushed back on AI responses to check and understand the accuracy of the response.

Automation bias can introduce significant risk, and legal education should address it. In fact, lawyers are ethically obligated to understand the risks of using technology. The ABA’s Model Rules of Professional Conduct also state in Comment 8 of Rule 1.1: “A lawyer should keep abreast of changes in the law and its practice, including the benefits and risks associated with relevant technology.”

Legal training has always been grounded in reasoning, and this expectation does not change just because AI is involved. Law schools should encourage students to use AI on assignments, provided they also require them to disclose that use and show they understand the output. They need to be able to explain why they used a certain case citation and specifically how it supports the position they are advocating.

This approach becomes a modern-day take on the Socratic method. It also reflects real-world questions lawyers get from the bench about the relevance of a case when arguing a motion or when they answer a client’s questions about why and how a given provision is included in a contract draft.

A student who uses AI to identify a case still needs to explain why that case applies. A student who uses AI to draft a contract still needs to explain why the language used works in the context of the contracting party’s need. These are core requirements of legal practice.

Some law schools are beginning to incorporate AI more directly. At Columbia Law School, a course taught by a practicing attorney allows students to build legal tools as part of the curriculum instead of submitting a traditional final paper. At Indiana University’s Maurer School of Law, professor Josh Kubicki offers multiple classes on artificial intelligence.

Such efforts reflect a broader shift, but they also highlight a larger challenge: AI is already changing how legal work gets done.

AI tools can improve efficiency, but they don’t necessarily contribute to the development of judgment or critical thinking, skills that have traditionally been built through repetition and direct engagement with the work.

AI changes that process. It can accelerate early drafts and reduce the time spent working through problems independently. That shift has implications for how skills develop, but it does not eliminate the need for them.

Legal education should reflect that reality. A more practical approach is to incorporate AI into the curriculum and hold students accountable for how they use it. Let them use the tools, then require them to explain their decisions, defend the output, and identify where it may be wrong or incomplete. These are the same expectations they will encounter in practice, so they must develop that muscle memory through repetition and direct engagement.

Law school is meant to teach people how to think. AI can generate an answer to provide knowledge, but the responsibility to understand and explain it remains with the lawyer and the law student alike.

See also:

Some law profs move to old-school assessment methods to combat AI

Artificial intelligence can be used for grading law school exams, but should it be?

Chad Main is an attorney and the founder of Percipient, a modern legal services provider. He is also the host of the podcast Technically Legal.

Mind Your Business is a series of columns written by lawyers, legal professionals and others within the legal industry. The purpose of these columns is to offer practical guidance for attorneys on how to run their practices, provide information about the latest trends in legal technology and how it can help lawyers work more efficiently, and strategies for building a thriving business.

Interested in contributing a column? Send a query to [email protected].