Look before you leap into AI, experts advise

DEPOSIT PHOTOS

They’re the two letters that won’t seem to go away, regardless of one’s profession: AI.

Through applications such as ChatGPT, artificial intelligence has been touted as everything from a timesaver to the thing that will kill all creative pursuits. But at this moment, do lawyers really have a solid use for it?

“My impression is that law firms understand that it’s important, but the implications aren’t quite clear yet,” said Eran Kahana, an attorney in Maslon’s Corporate & Securities Group. Kahana recently began teaching the class “Artificial Intelligence and the Law” as an adjunct professor at the University of Minnesota Law School.

“There’s a sense of urgency in law firms that this is something that they need to understand,” said Kahana. “Some are rushing to put applications in place. There are risks in doing that, but there’s a definite understanding that this is a technology that clients eventually will either expect to be used, or expect that it will be avoided, depending on their needs.”

AI’s potential is great, but as the saying goes, the scouts catch most of the arrows. A case in New York showed the risk of generative AI. In 2022, Robert Mata filed a personal injury action against airline Avianca Inc. After the case was removed to the Southern District of New York, Avianca filed a motion to dismiss. Mata’s response to the motion to dismiss was full of case law citations in his favor — and that also didn’t exist. They were generated by ChatGPT, the free “chatbot” that lets users refine conversations toward a desired length, format, style, level of detail and language.

“That case showed that these tools might not be as accurate as we want them to be,” said Kevin Frazier, an assistant professor of law at the St. Thomas University Benjamin L. Crump College of Law in Florida. Frazier was part of a University of Minnesota symposium on the topic earlier this month, “Artificial Intelligence (AI) in Legal Research and Decision-Making: Opportunities and Risks.”

At least one Twin Cities firm is putting some chips down on the burgeoning technology. Late last year Schwegman Lundberg & Woessner announced deployment of AI tools for use in its intellectual property practice. Founding partner Steve Lundberg stated at the time that the use of the tools was “a natural progression” for the firm and its enthusiasm for new technology.

Still, some say that when it comes to applying AI to your legal practice, a little deliberation might be in order. A recent survey by Lexis showed that about 15% of U.S. lawyers had actively used generative AI tools in their practice.

“Fifteen percent is greater than zero, but lawyers are tiptoeing toward adoption,” said Frazier. “I think that as we see more law firms finding the potential for improvement in the efficiency and quality of their work, there’s going to be greater financial pressure as well as client expectations that they use these tools.”

According to Minneapolis criminal defense attorney Ryan Garry, generative AI has the greatest potential in four areas: automation of legal processes, enhanced legal research, predictive analytics and case outcome assessment, and enhanced contract analysis.

But the potential of AI doesn’t mean it’s for every practice area, cautioned Kahana.

“I think litigation is an area where you might see more benefit,” he said. “I see less benefit on the transaction side.”

Kahana added that some AI-based applications, such as Lexis’s Lex Machina,  can be used for budgeting, litigation, or gaining insight on adversaries.

“It’s not the only application of its kind, but it’s one of the first ones,” he said.

At this point, according to Kahana, the technology is still requiring a significant amount of oversight by attorneys. That means the benefit of using AI “can be quickly eclipsed by the requirements imposed by state rules of professional responsibility to review and confirm that the output is accurate.

“Relinquishing control by the lawyer to the application is extremely problematic,” he said. “That causes a time commitment for review that can degrade any benefit of the application.

AI, according to Kahana, might be useful for brainstorming or for some rudimentary tasks. “It’s just a matter of time before those tools get better, but overall, these applications really aren’t all that good yet,” he said.

“I think every day, people are becoming aware of both its potential and its limitations,” said Frazier. “It seems inevitable that the use of AI in the legal field will accelerate.”