Skip to Content

Clinic Students Discuss the Evolving Role of AI at Conference

By Brooke Paz & Tristan Brown

What role should artificial intelligence play in clinical legal education? This question sparked a thoughtful discussion at this year’s Rocky Mountain West Clinical Conference in Denver, where representatives of ASU Law’s First Amendment Clinic presented on the opportunities and challenges of integrating AI into clinical practice.

Returning after a decade-long hiatus, the conference brought together clinical educators from across the region at the University of Denver Sturm College of Law to explore this year’s theme, “Creativity in Clinical Education.”

First Amendment Clinic Director Gregg Leslie, Fellow and Assistant Teaching Professor Martha Lemar, and third-year law students Tristan Brown and Brooke Paz led a breakout session entitled “Ready or Not, It’s Here: Clinical Education in the Time of AI Explosion.”

Putting AI to the Test

In preparation for the conference, Tristan and Brooke—under the guidance of Professor Lemar—tested AI’s capabilities by completing legal assignments independently, directing AI to complete the same assignments, and then using AI to review their original work. Then, they evaluated the tools’ strengths and shortcomings and reflected on their potential applications in future legal practice.

AI in Motion Drafting and Short-Form Informative Writing

Tristan drafted a Rule 12(c) motion for a judgment on the pleadings in a school speech case with two other student attorneys at the clinic. After the three researched this area and drafted the motion, he prompted ChatGPT (in a sandboxed environment) to draft a Rule 12(c) motion after providing it with PDFs of the three cases cited in the motion, the pleadings, the applicable local rule, and a description of the strategy behind the motion.

The most notable difference between the human-drafted motion and ChatGPT’s was length, as the latter was twice as long. ChatGPT did not fully grasp the nuances of the strategy underlying the motion or some of the requirements of the local rule. However, it was effective at pulling rule statements and other helpful language from the three cases to build arguments.

ChatGPT was most effective at identifying helpful admissions in the Answer, which aligns with how practitioners use AI (in sandboxed environments) to analyze documents such as medical records or claim files. Some of ChatGPT’s suggestions regarding admissions in the Answer that should be more prominently featured ultimately made their way into the actual motion. Overall, ChatGPT can save considerable time by extracting information from dense documents or assisting with revisions. However, using AI for tasks that require more strategic thinking, such as drafting a motion with a specific strategic objective or greater nuance, may result in important considerations being lost in translation.

To test this theory, Tristan repeated the experiment with ChatGPT but gave it a smaller, less intensive task: drafting an informative blog post on the First Amendment right to film police.  Here, when instructed to draft something short and informative, ChatGPT did a much better job of getting to the point and understanding the assignment than it did with the Rule 12(c) motion.

Tristan’s main takeaway from this experiment was that when AI is given a longer leash and less guidance, it can stray further from the original goal. When the parameters are narrower and the assignment is less variable, AI can save time on the more cumbersome aspects of legal practice, allowing attorneys to focus on areas where AI cannot replace them.

AI in Legal Research and Memorandum Writing

Brooke drafted a legal memorandum examining how public records laws apply to journalism programs at public universities. After completing her own research and analysis, she tasked both ChatGPT and Westlaw’s CoCounsel with drafting the same memorandum.

The difference in time commitment was striking. Researching, drafting, and editing the original memorandum took approximately 20 hours. The AI-generated memoranda took seconds to minutes to produce, followed by a few hours of review to assess their accuracy.

The differences went beyond speed.

ChatGPT produced a well-organized, largely accurate memorandum and proactively asked follow-up questions to refine its analysis. Although some wording and pinpoint citations required correction, the overall quality was impressive. Most notably, ChatGPT identified two useful sources Brooke had not uncovered during her own research, demonstrating AI’s potential to supplement independent legal research.

CoCounsel, on the other hand, included useful procedural recommendations and practical checklists for university journalism centers. However, its writing was convoluted and repetitive, it more than doubled the assigned page limit, and it omitted citations despite explicit instructions.

This experiment demonstrated AI’s potential to improve efficiency and research while reinforcing the need for human oversight to ensure accurate, reliable legal work.

Brooke also highlighted AI’s potential applications in client intake, mediation, communications, and technical troubleshooting. By streamlining administrative tasks and making better use of limited resources, AI could help law school clinics expand access to legal services in their communities.

Looking Ahead: AI and Professional Responsibility

Professor Leslie offered a broader perspective on AI’s implications for the legal profession. Drawing on insight from ASU Law Professor Sean Harrington, Director of the AI and Legal Technology Lab, Professor Leslie discussed how attorneys’ professional responsibilities may evolve as AI becomes increasingly integrated into legal practice.

While much of today’s conversation focuses on the risks of using AI, Professor Leslie suggested that failing to use it could eventually raise professional responsibility concerns. As AI tools become more sophisticated, lawyers who disregard technologies capable of improving the accuracy or thoroughness of their work may risk falling short of evolving standards of professional diligence. Such developments could even have implications for legal malpractice liability and insurance coverage.

Although these possibilities remain speculative, they raise an important question for clinical legal education: If tomorrow’s attorneys may be expected to use AI competently, what responsibility do law schools have to prepare students for that reality?

A Candid Conversation About AI in the Classroom

The session concluded with a spirited Q&A, as attendees raised concerns about AI’s practical and ethical implications, including whether reviewing students’ prompts and AI-generated work would increase faculty workloads.

While the panel acknowledged that supervising AI-assisted work presents additional challenges, the technology is already widely accessible to students. This leaves law schools with a choice to either teach students how to use AI responsibly or leave them to navigate its ethical boundaries alone.

Other attendees questioned how students might use AI without their professor’s knowledge. Tristan and Brooke described tools such as Google’s NotebookLM, which can help transform lecture and reading notes into podcasts, video explainers, flashcards, and practice questions. Far from replacing the learning process, these applications provide additional resources for students to review their materials, test comprehension, and untangle difficult concepts.

Brooke also emphasized that most students have strong incentives to use AI ethically, including comprehending the material to pass their exams and the bar, developing professional competence, and satisfying character and fitness requirements for admission to the profession. Using AI to evade learning, she suggested, would only undermine those goals.

Although not every concern could be resolved in this session, the discussion continued afterward, with several attendees approaching Tristan to learn more about the students’ experiences and insights.

As AI reshapes legal practice, clinics have an opportunity to prepare students to harness its benefits, navigate its risks, and exercise the independent judgment required to provide competent legal representation.

And yes, AI helped write this blog post.

Disclosure: Generative AI tools were used to improve grammar, spelling, and clarity. This blog post was drafted by humans, and all suggestions were reviewed, edited, and approved by the authors.