By: Alexa Chew, George R. Ward Term Professor and Clinical Professor of Law
With increasing frequency, the first question attorneys ask me when they find out that I teach legal writing to law students is this: “But what about AI?” Sometimes they ask questions driven by curiosity, such as “What are you doing about AI?” or “Are you teaching AI to students?” Other times they ask questions driven by panic, such as “Is AI going to ruin legal writing, legal practice, and the law as we know it?!???!??” Regardless of their line of questioning, what all of these folks want to know is how to prepare new lawyers to write in workplaces that increasingly rely on generative AI tools, whether that reliance is eagerly anticipated or begrudgingly tolerated. The future of AI in the law is uncertain, of course, but here are a few ways to think about how AI is currently affecting legal writing and might do so in the future.
One way to think about writing with AI tools is to view it as a form of delegation. Under the delegation model, instead of delegating a chunk of your own work to another human, you are delegating that chunk to the AI. To make this delegation easier to do (and perhaps easier for your soul to accept), remember that the principles of good delegation still apply. If you are delegating to AI, you must (1) define the task clearly, (2) match the task to the delegee’s skillset, and (3) review the delegee’s work rather than accepting it as perfect. Whether you are supervising a human or a machine, frustration will result if you fail to apply the first two principles (task definition, skillset matching), and ethical problems arise from not applying the last one (reviewing the work).
If you’ve ever been humbled by trying to delegate a familiar task, only to have that delegation end in abject failure, then you know firsthand that being able to do a task yourself does not automatically enable you to guide another person through that task. Delegation failure can happen when you don’t follow the three principles of good delegation. You need to know how to break the task into pieces, describe those pieces accurately, and recognize whether the pieces have been done correctly. If the task is a piece of writing, then effective delegation requires knowing how to talk about writing — how to talk about structure, legal authorities, paragraphing, tone, and so on. In other words, it requires a writerly vocabulary. A person might be an excellent legal writer and still lack the knowledge to delegate well.
So here is my professional secret that I can share with you: when I prepare to teach legal writers to work with generative AI tools, I start with teaching them how legal writing works. I give them a deep understanding of legal writing itself, for example, by building up a vocabulary to talk about legal writing. This deep understanding of legal writing will serve them well whether they’re doing their own writing, supervising other lawyers, or instructing a generative AI to outline the facts section of a brief. These skills are universal.
