Transform Legal Writing with AI: Challenges, Solutions, and Outlooks
Justice Scalia and Bryan A. Garner once posited: “[l]awyers possess only one tool to convey their thoughts: language.”[1] Large language models (LLM), such as those developed in recent years, have the potential to revolutionize this tool by enhancing clarity, efficiency, and accessibility in legal writing, allowing attorneys to focus more on strategic thinking and less on the mechanics of composition.[2]
However, the integration of AI into legal writing is not without its challenges, as lawyers must navigate concerns regarding the accuracy, ethical implications, and the potential for over-reliance on technology, which may ultimately dilute the understanding essential to effective legal practice.[3] Large language models have inherent drawbacks, one of them being “hallucination.”[4] This phenomenon, where AI-generated content may produce inaccurate or misleading information, underscores the necessity for lawyers to remain vigilant and critically evaluate the output of these tools before incorporating them into their work.[5] In a 2023 airline personal injury case, New York attorney Steven A. Schwartz improperly leveraged ChatGPT without realizing that the AI-generated content contained fabricated case law, leading to significant repercussions for his client and his professional reputation.[6] This incident serves as a cautionary tale, highlighting the critical importance of thorough verification and the need for attorneys to maintain a strong grasp of the legal principles at play, ensuring that AI serves as an aid rather than a substitute for the lawyer’s expertise and judgment.[7] AI drawbacks such as “hallucination” could also be the underlying reason contributing to the slow adoption of AI among law firms.[8]
One potential safeguard for integrating AI into legal writing is to ensure that lawyers remain at the helm of the writing process. Rather than relying on ChatGPT to draft a legal memo from scratch, lawyers should continue to craft their own documents and utilize AI primarily for more mechanical tasks.[9] This approach not only mitigates the risks associated with inaccuracies but also ensures that the attorney’s voice and legal reasoning remain at the forefront of the document, reinforcing the essential human element in legal communication that AI cannot replicate. Furthermore, it is crucial for law firms to invest in training and developing a clear framework for the ethical use of AI tools within their practices, fostering a culture of collaboration between technology and legal expertise.[10]
AI technologies such as retrieval augmented generation (RAG) can be particularly beneficial in this regard, as they allow lawyers to control what information AI is leveraging in writing a particular content.[11] By leveraging RAG, attorneys can ensure that the AI draws from verified and relevant legal resources supplied by attorneys themselves, thus enhancing the accuracy of the generated content while maintaining the integrity of their arguments and analyses.[12] This controlled approach not only mitigates the risk of error but also empowers lawyers to remain intimately involved in the crafting of legal documents, ensuring that their professional insights and contextual understanding are seamlessly integrated into the final legal documents. Below is an example of leveraging RAG to retrieve specific legal arguments in the Dobbs dissent opinion filed by Justice Breyer, Justice Sotomayor, and Justice Kagan.[13]
Source: Ray Sang, RAG Demonstration Using Dobbs Opinion, performed Nov. 30, 2024, using https://www.inkwise.ai.
No matter how people use AI, one thing we can agree on is that AI can help us write.[14] One of the major benefits of AI is that it dissipates the difficulty of writing.[15] It is my prediction that in the future, there will be a polarizing force in legal writing due to the use of AI—lawyers who can use AI properly will be among the best advocates for their clients, and lawyers who can’t struggle to keep pace in an increasingly competitive legal landscape. There will be few lawyers in between.
We conclude with the wise writing from Judge Castel in his ChatGPT sanction decision: “[t]echnological advances are commonplace and there is nothing inherently improper about using a reliable artificial intelligence tool for assistance. But existing rules impose a gatekeeping role on attorneys to ensure the accuracy of their filings.”[16]
Footnotes
