MIT professor allays bar's fears of 'robot lawyers'

By Pat Murphy
BridgeTower Media Newswires
 
BOSTON — Frank Levy doesn’t see robots replacing lawyers anytime soon, but the MIT professor refuses to hazard a guess as to how artificial intelligence will have transformed the legal profession 50 years from now.

Only about 13 percent of all the work done by corporate lawyers today could be automated, even if firms were to immediately maximize their use of existing technologies, Levy says.

“Because these technologies are not going to be implemented all at once, the more realistic assumption is that computers would be able to absorb about 2 percent [of the work of lawyers] a year,” he adds.

Those are some of the key conclusions of “Can Robots Be Lawyers?” a research paper authored by Levy and Dana Remus, a professor at the University of North Carolina School of Law.

Levy and Remus originally posted their paper for comment on the Social Science Research Network in December 2015. After some minor revisions, it is set to be published in the Georgetown Journal of Legal Ethics.

Levy underscores the point that he and his co-author focus their analysis on the work performed for corporate clients. He acknowledges that the study does not look at how online legal tech companies such as LegalZoom have eaten into the preparation of basic legal documents like simple wills or for the incorporation of a small business, once staples of the sole practitioner.

He expects solos will continue to be vulnerable to the loss of work related to the preparation of simple, routine forms and agreements.

On the other hand, Levy sees a sort of natural firewall when it comes to litigation, a firewall that technology has so far largely been unable to penetrate.

“If it’s non-adversarial, like the incorporation of a small business, services like LegalZoom can really simplify things and in some ways remove the need for a lawyer,” he says. “But if it’s adversarial, templates and so forth go out the window.”

He’s hesitant to say if or when computers will advance to the point of being able to make the close judgment calls that providing legal advice entails.

“There are certain really complicated decisions involving applying a whole lot of experience and other factors that are going to be hard to put into a computer model,” he says.

However, Levy points to recent leaps with respect to e-discovery as foreshadowing the course of progress in the future.

“Computers are doing a better job of understanding language and that is what has opened up the possibilities of automation,” Levy says.

Ten years ago, apart from searching for some key words, computers couldn’t deal with complex protocols for identifying particular documents relevant to a discovery request or order, he says, noting that the advent of machine learning in computer science has changed that.

“Suddenly you could take advantage of the fact that [e-discovery] was a routine job,” Levy says. “There are going to be other developments like that where you get better ways of representing language or better ways of searching that’s going to open up more of these areas that have been protected from computerization up until now.”

For the near term, Levy sees those technological advances as further automating the document construction and editing process for lawyers. But he remains leery about predicting that revolutionary advances in artificial intelligence will have robots dispensing legal advice 50 years into the future.

“A lot of these long-run projections on the future of work only talk about the technology, like pushback from institutions in society doesn’t exist,” he says. “You’re leaving out a big part of the story. How are people going to react? What do the courts do?”