Artificially Intelligent Legal Research

May 24th, 2016 / By Deborah J. Merritt

At least three law firms have now adopted ROSS, an artificial legal intelligence system based on IBM’s pathbreaking Watson technology. The firms include two legal giants, Latham & Watkins and BakerHostetler, along with the Wisconsin firm vonBriesen. Commitments by these firms seem likely to spur interest among their competitors. Watch for ROSS and other forms of legal AI to spread over the next few years.

What is ROSS, what does it do, and what does it mean for lawyers and legal educators? Here are a few preliminary thoughts.

The ROSS Platform

I wish I knew enough about computer programming to explain exactly how ROSS operates. I don’t really understand the inner workings of “natural language processing” and “machine learning.” But I watched Watson’s performance on Jeopardy five years ago, and I could see that Watson’s abilities stretched far beyond Google, Westlaw, or any other computer-based research program I currently use.

Sure, Watson made a few goofs–but that’s what humans are for. We can laugh when a computer gets confused and gently correct the program. Overall, Watson found the right answer more often, and more rapidly, than its human competitors.

Since that famous Jeopardy performance, Watson has been studying law, medicine, and other subjects. Its first venture into legal research, through the ROSS program, focuses on bankruptcy law. ROSS, quite simply, can identify and analyze bankruptcy-related statutes, rules, and case law more quickly and accurately than a team of lawyers.

ROSS doesn’t eliminate lawyers: A lawyer must still interview a client, assess the client’s needs, and ask ROSS about the relevant law. But ROSS helps the lawyer identify that law much more efficiently than other research tools.

For now, ROSS works only on bankruptcy problems. Its developers, however, plan to branch into other areas of law. ROSS (or his cousins) may soon advise you on intellectual property, securities, and other areas of legal expertise.

Impact on the Profession

ROSS’s creators are quick to reassure lawyers that the software won’t replace lawyers. Literally, that’s true: the program won’t eliminate the need for human attorneys. It seems virtually certain, however, that tools like ROSS will further reduce the number of lawyers that clients need.

If an associate can find the relevant case law in an hour, rather than a day, then firms won’t need as many associates. If in-house lawyers can find that law just as quickly, they won’t assign as much work to firms. As programs like ROSS develop, they further constrain lawyer employment.

At the same time, ROSS will increase demand for the skills that people perform better than machines. Law schools often downplay “soft skills” like interviewing, counseling, and negotiation. These skills, however, have a human element that machines can’t match. Interviewing isn’t just a matter of asking the right questions; the best interviewers establish trust and rapport with their subjects. The same is true of successful negotiators and counselors.

The very humanness of these skills deceives us into thinking that they can’t be taught. How do you teach someone to establish rapport with a client, witness, judge, or opposing counsel? That seems like teaching someone to be human! Surely this isn’t something that law schools can or should teach.

But if we don’t begin studying and teaching these skills, our graduates will lose their competitive edge. Lawyers need to be hard-headed and logical, if only to keep up with the machines. But we also need to cultivate the talents that machines can’t reproduce–and that clients will be willing to purchase.

Hope for the Justice Gap?

If applied to fields like family, criminal, and consumer law, ROSS could greatly improve legal services for low- and middle-income clients. In legal representation, money buys time. A client who can afford only an hour or two of legal representation gets limited help for that price. For public defenders and legal aid lawyers, chronic underfunding leads to high caseloads and time-crunched representation.

ROSS has the potential to make the most relevant law immediately available to lawyers. We could harness its power to develop state-by-state knowledge banks about laws that affect low- and middle-income clients.

Clients might access some of that legal knowledge on their own, but ROSS’s real impact would be enabling lawyers to work faster for their low- and middle-income clients. Imagine ROSS working for an agency like Chicago’s Coordinated Advice and Referral Program for Legal Services. That organization has already developed human staffing models for delivering legal services efficiently. Adding ROSS to the equation could multiply the agency’s reach.

For that to happen, of course, ROSS’s power has to be available at low cost for organizations that serve low- and middle-income clients. Otherwise, ROSS will become yet another tool that widens the gap between rich and poor clients. Rich clients and their attorneys will be able to afford ROSS’s super research skills, while less fortunate clients work with older technologies.

Let’s begin thinking about ways to integrate ROSS into law practice, use its capabilities to complement human talents, and extend the benefits to all users of the legal system.