•  
  •  
 

Abstract

Much has been made in recent years of the capabilities of generative artificial intelligence ("Generative AI") programs in performing law-related tasks. We have learned, for example, that OpenAI's product, GPT-4, is capable of passing the Uniform Bar Exam with flying colors. ChatGPT also earned passing grades on the final exams in four different law school courses. Another study found that "AI assistance consistently induced large declines in the amount of time taken to complete tasks" like drafting complaints, contracts, and employee handbook sections. These and other success stories have led commentators to predict that Generative AI programs "will revolutionize what it means to draft legal documents." We have also learned about Generative AI's limitations and the potential for its misuse. In 2023, a lawyer for the plaintiff in a personal injury suit used ChatGPT to draft a brief in opposition to a motion to dismiss. The brief, which the attorney filed in a federal court in New York, relied on six cases that did not exist. In July 2023, a lawyer in Colorado was fired by his law firm after using ChatGPT to write a brief that cited nonexistent cases. In November 2024, an attorney in Texas was fined two thousand dollars and ordered to take a continuing legal education course on Generative AI after he filed a brief that referenced nonexistent cases and quotations. In 2025, two federal district court judges withdrew orders after attorneys in the cases pointed out that the orders named plaintiffs who were not parties to the case, referred to non-existent cases, misstated the outcomes of other cases, and contained fake quotations. Clearly, the legal profession needs a better understanding of what these programs can and cannot do. To that end, this Article analyzes Generative AI in the context of legal writing. Specifically, I wanted to know how Generative AI would perform on assignments that law students typically complete in their first-year legal writing courses. While this Article focuses primarily on assignments given in law school, the question of Generative AI's capabilities in the realm of legal writing has profound implications for the practice of law. Communicating effectively in writing is a foundational lawyering skill. If Generative AI can improve the quality of legal documents, then lawyers everywhere should take notice. Recent developments suggest that AI is here to stay and that its use and capabilities will only expand in the years to come. In January 2025, President Donald Trump announced that the federal government would invest up to $500 billion to fund infrastructure for artificial intelligence. This investment will enable private technology companies to build data centers necessary to power AI programs. As AI becomes more prevalent, lawyers will increasingly rely on it to better serve their clients. Therefore, it is critical that we understand the benefits and limitations of this new technology as applied to the practice of law. There are three parts to this Article. Section I provides background on Generative AI and summarizes the available research on its ability to perform legal writing tasks. Section II describes simulations in which I used ChatGPT to write the Fall Final Memo and Spring Final Motion Memo assessments assigned to my students in the 2023–24 academic year. It goes through different components of each assignment, identifying some areas of legal writing where ChatGPT excels, and several more where it struggles. Finally, Section III considers how legal writing professors and practicing attorneys should use Generative AI, in light of the strengths and weaknesses identified in Section II.

First Page

69

Download

Share

COinS