AI is here to stay: Law profs highlight benefits, risks for attorneys

Listen to this article

According to a study done in 2024, the use of AI by legal professionals jumped from under 20% in 2023 to nearly 80% in 2024. That number is likely higher this year. With AI becoming part of the practice of law, lawyers have questions about its use and its limits. Two law professors with expertise in the use of AI and legal practice address some of these questions.

Help lawyers to be better writers

Shaun Jamison, associate dean of faculty at Purdue Global Law School, who is based in Minnesota, spoke to how generative AI can be used to help lawyers write better. He also noted the uses of AI beyond simply drafting briefs or document review. It can be used to draft schedules, or even letters to opposing counsel. He also says that it can, perhaps, help us relate better to people by making work more palatable, on top of being better organized.

“It’s odd, because AI can’t feel, but it understands the language of feeling,” Jamison said.

Generative AI users can, for instance, put in text and ask for opinions on the tone, and AI can provide alternatives that might be received better.

There is the stigma of sounding “ChatGPT-ish” — essentially for writing to have a form that is associated with the use of generative AI. Some “red flags” have been the use of em dashes, the word “consequently” or overly generic statements.

Jamison says that, just because writing might have these flags, it does not necessarily mean that ChatGPT was used to draft the writing.

“I would caution people accusing people or insinuating that people are using ChatGPT based on tells because ChatGPT and Gemini and all those were trained on the sum of people’s writing,” Jamison noted. “It’s kind of regurgitating what it saw.”

There also are ways, Jamison notes, to not sound like you have used generative AI. One way to not sound ChatGPT-ish, Jamison says, is simply to not have ChatGPT write for you.

“You can ask it to ‘identify the issues that I should write about’ or ‘write an outline for me on this topic’ and then you write it,” Jamison said. “Then, it won’t sound like AI, but you will still need to see if AI left anything out. He also notes that, if you do have AI draft your work, that you can always edit it.”

“The bottom-line advice I give folks is to only use AI when you are confident in your ability to evaluate the output,” said Daniel Schwarcz, Fredrikson & Byron Professor of Law at the University of Minnesota Law School.

“You can also ask it not to sound like ChatGPT,” Jamison notes. “You can tell it to ‘make this sound more personable’ or ‘make this sound like a person with English as a first language.’ You can trick it into sounding better.”

Communicating about AI usage with clients

Now that lawyers are regularly using AI, firms are considering best practices. One question is whether, and to what extent, clients should know about a lawyer’s use of AI.

“There are pretty well-known risks that come with AI use, as well as potential benefits, and it’s something on which consent just exists currently in the legal profession,” Schwarcz cautioned. “Given that, I would say that disclosure is appropriate.”

“Clients want results, but they also want transparency,” Jamison agreed. “You should have a policy that your clients are made aware of as to how you will use AI and how you will protect their interests when doing it. How are you protecting their privacy?”

He recommends telling clients, “We use AI this way. We train our lawyers in this fashion. This is how we protect your interests and make sure the work is accurate.”

As a practical matter, Jamison suggests stripping out any client identifying information before asking AI a question.

Lesser-known concerns regarding AI

Stories of lawyers submitting AI-drafted briefs that have errors and false cases cited have made headlines. While these risks are present, Schwarcz says that these are far from the biggest concerns with the technology.

“The clearest, most obvious risks get the most attention because they are the most obvious, and that’s hallucinations,” Schwarcz asserts. “To be honest, I tend to think that those are overhyped because they are so obvious when they occur—and they get a lot of attention—but they’ve very easy to avoid, for the most part.”

“There are much more significant risks that are less well-understood that aren’t always fully disclosed,” Schwarcz explained. “For me, the biggest one is that, if you’re using an AI system, there’s always the risk that you are going to be missing certain issues, or not seeing certain arguments that you would otherwise see. There is a hidden risk of degradation in quality of representation.”

Jamison also noted that environmental concerns about the amount of energy and water that data centers use are underappreciated.

“It’s exponentially higher when you are using AI,” Jamison said. “My suggestion is to think about what the best tool is to accomplish what you need.”

“If you can crack a book, do it,” Jamison said. “If you don’t need to use AI, don’t, because it does use a lot of energy and water.”

The future of AI and law

It is clear that the use of AI is a part of legal practice, and tomorrow’s lawyers will not know legal practice before AI. It is unknown to what extent AI will replace the tasks that lawyers do, and how it will change the lawyer’s role. Jamison suggests that, by outsourcing more of the mundane tasks, lawyers can focus on actual lawyering. This could bring more value to clients.

If AI could eventually replace lawyers (or significantly change the skills of the job), could it replace judges?

“If you are talking about where the technology is today, you probably don’t want it deciding court cases. It’s not ready for that,” Jamison said. “Could it be at some point with the proper supervision? I think, yeah.”