By Gwyneth K. Shaw
As AI increasingly shapes legal practice, Berkeley Law has a new course starting this fall aimed at educating students about the technology’s strengths and weaknesses while giving them more than 100 hours of hands-on training.
Taught by Berkeley Center for Law & Technology (BCLT) Executive Director Wayne Stacy — a former Big Law litigator and U.S. Patent Office leader — AI and the Practice of Law: Public Interest & Private Practice will focus on enhancing legal skills without lowering the threshold for human judgment and input.
“This is the fundamental question: How are you going to do higher quality legal work with AI than you’re going to do by yourself?” Stacy asks. “It’s not just being faster and cheaper. The jump in the quality of work, in the creativity for your client, is the difference. That’s what I want my students to learn.”
Stacy, who has a background in computer engineering and took the helm at BCLT in 2021, started as something of a skeptic. After reading a stream of articles predicting AI would make young lawyers essentially obsolete, “it seemed like scare tactics to get people to click on headlines,” he says.
Sure, he reasoned, there are some routine tasks that AI could take over. But the law is full of complexities that a model might miss.
Like any good patent lawyer, Stacy started teaching himself the technology. Quickly, he came to understand that while AI models are continuously improving, it’s the platforms set up for lawyers to use them that are going to change the legal industry. Once someone understands the right way to use the platform, future innovations are built in.
“I realized that what we needed is a class to show what the foundations of an AI-based legal platform are and how you use it,” he says.
Cracking the code
To develop the three-credit course, which is open to 2L and 3L students, Stacy talked to more than 20 Big Law lawyers about how their firms use AI, and what’s working and what’s causing problems. He also worked closely with Berkeley Law Librarian Kristie Chamorro, who has been studying these tools for several years and is assisting with the course.
He settled on Anthropic’s Claude for the course, because he feels it’s the most transparent and easy to understand platform. But the course is designed to be “platform agnostic,” likening the skill-switching abilities to that of an iPhone user picking up an Android phone. Once Stacy learned what the law firms were struggling with, he started digging into why they weren’t succeeding.
“I had to do some diagnosis as to why they were failing,” he says. “Were they not accessing the right subscription? Did someone forget to allow web searching as part of their setup?”
As he decoded the issues, Stacy found optimal schemes that he plans to use to train students.
“The great thing about this class is we’re designing something that’s not trailing what law firms are doing — we’re teaching something that’s way ahead of what firms are doing,” he says. “I can tell students, ‘the law firm you go to may not have this yet, but you’ll know how to handle it when they catch up.’”
Chamorro says the course is “a terrific addition to the curriculum.”
“Critical literacy around AI is now part of being practice-ready,” she adds. “Students need to develop an understanding of what these tools can do, where they fall short, and how to participate confidently in conversations about their use.”
Judicious deployment
Part of the challenge for lawyers, Stacy says, is determining when and where to use AI. He divides the current playing field into three categories:
- “AI automated” work: Low-complexity tasks like document review or basic research queries. The risk to not using AI is that clients are unlikely to pay attorney hourly rates for these tasks moving forward.
- “The disruption zone”: Mid-complexity work, the bread and butter of a law firm, is being impacted. AI can do some of these tasks, but judgment and oversight are critical. Lawyers who master this space, Stacy says, will dramatically outperform their peers.
- “AI augmented” lawyering: Offering clients a better version of representation, from novel arguments to strategic advice, requires human judgment and careful use of AI. But fluent lawyers can practice at a significantly higher level than their peers.
Where and when lawyers use AI to enhance their work depends a lot on what they’re doing, Stacy says. For example, personal injury plaintiff’s lawyers are heavy adopters because many of their cases resemble one another, at least in particular stages, so using AI can help them handle more cases and potentially even bring down the threshold for taking on a client.
For transactional lawyers, on the other hand, using AI really means creating workflows that make a team both more efficient and able to analyze documents and cases in a way that highlights things a human-only team might have missed.
“The workflows are really the secret sauce for any lawyer or group of lawyers,” Stacy says. “You can standardize a process and make sure you’re getting the same quality of analysis and escalating the things that need to be escalated. It’s obviously going to be faster but it should also be better-quality work.”
The course’s 11 assignments will teach a wide range of skills, starting with basic prompting and ascending to tools that use Claude’s features in concert to do research, draft contracts, and reach toward creative client solutions.
A strong foundation
Stacy knows AI is going to be influential in the legal field. But lawyers themselves, he insists, will remain the most important component of representing a client.
“What happens when every lawyer has access to the same platform and the same data set? What distinguishes one lawyer from the next?” he asks. “The answer is judgment, and we need to build that first.”
The need for students to build a foundational understanding of the law before they start trying to enhance those skills with AI is why the course isn’t open to 1Ls, Stacy adds. With an eye on ensuring students get that grounding, Berkeley Law’s faculty recently adopted an AI policy that defaults to not allowing students to use it, although instructors may allow elements as they see fit.
There are just some courses, particularly the black-letter subjects taken during the 1L year, where AI doesn’t belong, Stacy says.
“We know that the foundational legal research and writing courses work. The risk of cognitive offload by putting AI in those courses is too high, and there’s really no benefit to it,” he says. “You need to learn these skills, and I don’t think you can elevate them with AI while you’re still learning them.”
Assistant Dean of Career Development Eric Stern says the efforts of the school’s Legal Research, Analysis, and Writing Program faculty and the law librarians to stay on top of the latest developments and courses like Stacy’s are both reasons why he’s certain students will continue to be exceptionally prepared for practice.
“Berkeley Law is a forward-thinking, innovative and intellectually rich community that strives to teach students not just how to think like lawyers, but also how to approach the practice of law through a socially conscientious lens,” he says. “I am confident that our faculty and staff will equip students with the career readiness skills they will need to ethically and responsibly utilize legal AI tools to enhance their practice and contribute to the mission of their employers no matter what kind of work they are doing, and we will do this without compromising our responsibility and commitment to teach students how to think, write, and analyze as lawyers.”
Berkeley Law students come to the school to get a top-notch legal education, Stacy says. And while Claude might be able to pass law school, it will produce average quality work.
“If you use it correctly, you really can do augmented legal analysis. It can be a real add-on if you’ve got the right kind of work and you use it correctly,” he says. “We’re trying to put lawyers at the top into the profession — to be change agents and real thought leaders.
“People don’t come here to be average, and that’s the best AI is going to get you to if you use it incorrectly.”