Onto the green, with luck, your science takes you. But when it comes to putting you will work by art and hunch. -- Columbia Law School Professor Karl Llewellyn, 1930
As a lawyer working on the team that builds ROSS the question I’m asked most often is whether artificial intelligence threatens to minimize the value of the lawyer’s judgment. After twenty years of practicing law myself I understand the concern. Legal skill is hard-won through trial and error, courtroom wins and--most importantly--tough courtroom losses. If you haven’t pulled the arrows out of your back more than a few times, you just can’t understand how hard legal experience is to earn. Seasoned lawyers don’t worry about losing their jobs to AI, they worry that technology will drive a race to the bottom; that the pressure to lower costs will diminish the value of experience.
Artificial Intelligence Combines Art and Hunch
All the features we’ve built into ROSS are designed to use the science of machine learning to maximize the value of experience. The algorithms ROSS uses to retrieve and rank cases are tuned to respond to each query with a set of curated decisions. The generate overview and find similar language functions deepen your understanding of how those cases are grouped and helps you explore additional authority. ROSS doesn’t spit out answers, it returns support for the best argument the lawyer can craft. Working every day with some of the best machine learning scientists in the world, I’m confident that ROSS will continue to provide better and better legal research results but never diminish the value of lawyers’ reasoning.
Still, lawyers are asking the right questions about the balance between science and art in our profession. And I haven’t found a better set of answers than the nearly 90 year-old lectures by Professor Karl Llewellyn collected in The Bramble Bush. The lectures were written as an introduction to the law and legal study for first year students at Columbia Law School in the 1920s. I’ll forgive you if you skimmed Professor Llewllyn’s advice when you were in law school. His language is dense for modern readers. But despite the slow going in some of the lectures, I recommend The Bramble Bush to anyone who is thinking critically about the intersection between law and technology.
Llewellynists Through and Through
One of Professor Llewellyn’s foundational positions is that an individual judicial decision cannot have meaning by itself. Standing alone, he reminds us, a case gives no guidance. “What counts,” he wrote, “what gives you leads, what gives you sureness, that is the background of the other cases in relation to which you must read the one.” According to Llewellyn, the prior decisions provide context and meaning for the words and technical terms used in the case you’re reading. “But above all, they give you the wherewithal to find which of the facts are significant, and in what aspect they are significant, and how far the rules laid down are to be trusted.”
From that premise--and nearly one hundred years ahead of big data and fast processors--Professor Llewellyn unknowingly laid out the strengths and weaknesses of the algorithmic approach to legal analysis:
“In the first case you have facts a and b and c, procedural set-up m and outcome x. In the second case you have, if you are lucky, procedural set-up m again, but this time with facts a and b and outcome y. How now, are you to know with any certainty whether the changed result is due in the second instance the absence of fact c or to the presence of the new fact d? The court may tell you. But I repeat: your object is to test the telling of the court. You turn to your third case. Here once more is the outcome x, and the facts are b, c and e; but fact a is missing, and the procedural set-up this time is not m but n. This strengthens somewhat your suspicion that fact c is the lad who works the changed result. But an experimentum crucis is still lacking. Cases in life are not made to our hand. A scientific approach to prediction we may have, and we may use it as far as our materials will permit. An exact science in result we have not now. Carry this in your minds: a scientific approach, no more. Onto the green, with luck, your science takes you. But when it comes to putting you will work by art and hunch.
I’m going to find a portrait of Professor Llewlleyn and hang it on our wall here at ROSS (if you have one, send it along). For we’re Llewellynists through and through.
We continuously refine ROSS’s fact-finding models to help lawyers respond to “absence of fact c or to the presence of the new fact d.” No legal research platform provides a more thorough analysis of procedural posture than ROSS. That’s because our team of lawyers understood from the beginning that you have to match “procedural set-up m” to arrive at “outcome x.” And we’re always working under “the suspicion that fact c is the lad who works the changed result.” (The vivid image of a fact working secretly to change the results of a case is my favorite image in the The Bramble Bush.)
An Exact Science We Have Not Now. Probably Never.
Ultimately, our work is driven by a profound respect for the fact that cases in life are not made to our hand. Nearly 100 years later, Professor Llewellyn is still correct that an exact science we have not now. It’s unlikely that he’ll be proved wrong anytime soon. It is hard to imagine a day when a machine will value and weigh equity and empathy and justice the we do as lawyers and judges.
I came away from The Bramble Bush convinced that Professor Llewellyn would be a champion for technology used in the service of legal reasoning. His lectures are pragmatic and forward-thinking. He worried that inefficiencies in the practice of law sometimes worked to the detriment of clients. Mostly, he returned again and again to the theme that the law is constantly changing and lawyers must adapt to change to succeed. At ROSS, we’ve taken Professor Llewellyn’s lessons to heart.
Charlie von Simson is a legal subject matter expert at ROSS. He practiced law for twenty years before running away to join a startup.