How Artificial Intelligence Is Transforming Legal Research

By DAVID LAT

Technology and innovation are transforming the legal profession in manifold ways. According to Professor Richard Susskind, author of The Future of Law, “Looking 30 years ahead, I think it unimaginable that our legal systems will not undergo vast change.” Indeed, this revolution is already underway – and to serve their clients effectively and ethically, law firms must adapt to these changing realities.

One thing that remains unchanged, however, is the importance of legal research. In the words of Don MacLeod, Manager of Knowledge Management at Debevoise & Plimpton and author of How to Find Out Anything and The Internet Guide for the Legal Researcher:

As lawyers, you need to be on top of the current legal landscape. Legal research will allow you to advise your client on the standards of the law at this moment, whether they come from case law, statutes, or regulations.

The importance of legal research persists, but how it’s conducted is constantly advancing and evolving. Just as attorneys who used hard-copy books for all of their legal research would be amazed by online legal research services like Westlaw, attorneys using current services will be amazed by the research tools of tomorrow, powered by artificial intelligence and analytics.

For a case study in how AI and analytics are already transforming legal research, consider Westlaw Edge – the just launched, reimagined version of Westlaw, the industry-leading online legal research service from Thomson Reuters.

A BRIEF HISTORY OF LEGAL RESEARCH

The history of legal research is the history of recorded law. The earliest laws were not recorded, a source of great unfairness to the people governed by them. For example, the Twelve Tables – the earliest written expression of ancient Roman law, dating back to around 451-450 BC – came about at the insistence of plebeians who felt victimized by inconsistent judgments based on unwritten custom.

The most famous example of ancient written law is the Tables of the Law or the Tablets of Stone – aka the Ten Commandments. According to the Bible, these two pieces of stone were inscribed with the Commandments when Moses ascended Mount Sinai, as written in the Book of Exodus. Stones inscribed with the Ten Commandments date at least as far back as the late Roman-Byzantine era (300-640 AD).

 

Jumping to this side of the Atlantic Ocean and focusing on American law, the earliest case reporters date back to 1789. But these first reporters were very different from the ones we know today. As Syracuse Law professor Ian Gallacher explains, “Early volumes of court decisions were generated by reporters from their own notes and impressions of court proceedings rather than reflecting an official record of proceedings.” Official reporters did not get appointed until the early- to mid-1800s, depending on the jurisdiction.

And even after the advent of official reporters, Gallacher continues, “legal research as a discipline was unknown; lawyers had to read all the opinions a court issued in order to know what the law in that particular jurisdiction was, and they had to annotate those opinions to keep themselves up-to-date.” In other words, early case reporters contained nothing but the cases, without any analysis, organization, or enhancements like summaries or headnotes.

Enter John B. West, a businessman from Massachusetts who moved to St. Paul, Minnesota, where he took a job as a traveling salesman. West was not a lawyer, but many of his customers were. In 1872, he quit his sales job and became a publisher of law books, starting the business that would eventually grow into the West Publishing Company.

John West began by publishing compilations of opinions from courts in several northwestern states (which eventually became the North Western Reporter, the first of the West regional reporters). In 1897, he started issuing reporters that contained summaries of the holdings of cases, as well as West’s now-iconic “key number system” – a powerful indexing tool that revolutionized legal research and remains foundational today.

Over the subsequent decades, case reporters were joined by a wealth of secondary sources, including treatises, Restatements, and various legal periodicals. Case reporters themselves came to be supplemented with “pocket parts,” pamphlets inserted into special pockets built into the back covers of reporters that would update the cases contained in the reporter. (Lawyers of a certain vintage will remember pocket parts, perhaps with nostalgia or perhaps with dread.)

But for the most part, as Don MacLeod of Debevoise explains, “Things didn’t change all that much for the longest time. The law was found in books, housed on dusty old shelves in your law school, law firm, or county law library. That’s where your knowledge was kept — that and in the heads of more experienced lawyers. Junior associates would have to head to the library, park themselves in carrells, walk through the books, and report back to the partners.”

Then, as MacLeod puts it, “the asteroid struck” – namely, the arrival of computerized legal databases like Westlaw and LexisNexis, where cases were continuously published, updated, and annotated. These services launched in the 1970s and grew dramatically in popularity during the personal-computer revolution of the 1980s.

Online legal research services, now made available over the internet (as opposed to dedicated terminals), today dominate the landscape. Law students, young lawyers, and not-so-young lawyers conduct the vast majority of their legal research online. Camping out at the library is no longer required.

“Today we can get our hands on a lot of information and have it analyzed quickly and powerfully, which wasn’t possible in the past,” MacLeod says. “And now we have 24/7, mobile access to information, wherever we are.”

Online legal research services have transformed the work of lawyers, to an extent that the customers of John B. West could never have imagined. It is fair to say that today, legal research as a whole is faster, easier, and more accurate than ever. But powerful new tools bring new challenges as well as opportunities, to which we now turn.

THE CHALLENGES OF LEGAL RESEARCH TODAY

Far from being a rote exercise, legal research has always been a vital and demanding enterprise, both an art and a science in different respects. In the words of Justice Felix Frankfurter:

[R]esearch requires the poetic quality of the imagination that sees significance and relation where others are indifferent or find unrelatedness; the synthetic quality of fusing items theretofore in isolation; above all the prophetic quality of piercing the future by knowing what questions to put and what direction to give inquiry.

Frankfurter’s description of the challenge nature of legal research remains true today, despite all the new technologies available to us.

“There have been great advances during my lifetime in terms of what legal research can do,” says Leann Blanchfield, Vice President of Content Strategy and Editorial at Thomson Reuters. “But the requirements of being a lawyer have gotten much harder.”

Think of legal research like a jigsaw puzzle, but with some added complications. You don’t know the picture the completed puzzle is supposed to form, you don’t know how many pieces the puzzle has, and you have to assemble the puzzle under extreme time pressure. This analogy captures the two great challenges faced by lawyers performing legal research today: they don’t have enough time for the task, and they lack confidence in their results.

On the issue of time, the pressures facing lawyers are more severe than ever. For lawyers on the billable hour, time is literally money – and the rise of alternative fee arrangements makes lawyer time even more precious. If lawyers are working on a fixed or flat-fee arrangement, how much time they spend on research could make or break their success.

One aspect of legal research that eats into lawyer time is converting the raw data produced by research into usable work product. Part of the difficulty here is that current research solutions often produce results that are far removed from what the lawyer is actually looking for. As explained by Khalid Al-Kofahi, Vice President of Research and Development at Thomson Reuters:


Another feature of today’s research environment that contributes to the time crunch is the dramatic increase in the amount of available information – not just the ever-increasing amount of law on the books, but also all of the secondary sources that could be helpful to attorneys.

This explosion in information contributes to the second major problem facing legal researchers – the crisis of confidence when it comes to the accuracy and reliability of research results. As Leann Blanchfield notes, “In light of everything that’s out there, how can you have the confidence that you have found everything you are looking for?”

“Today, as a law librarian, I need to explain the dangers of relying too much on wonderful modern technological tools like Google that seem to deliver answers magically,” says Don MacLeod of Debevoise. “Those answers might not be the right answer or the thorough answer. The ease of use also lulls researchers into asking simplistic questions that can be asked using Google syntax and not the more powerful and queries that can be created using modern legal search tools. Google is a good place to start your research, but a terrible place to end it.”

The complexity of law and the legal system today also contributes to a lack of confidence in another sense – confidence in terms of what judges will do with the research and arguments presented to them. How can lawyers have any comfort level about how a specific court or other decision-maker will respond to a given claim or argument? Traditionally lawyers have made these assessments and advised clients based on their “experience” – which critics might say is little more than guesswork.

So lawyers doing legal research today face the challenges of time pressure and lack of confidence – and, unfortunately, there’s an unavoidable tension between addressing both of them.

If lawyers want greater confidence in their results, they need to spend more time on legal research – time they simply don’t have. “Lawyers desperately want to save time, but they are terrified that they will miss something,” observes Tonya Custis, a Research Director at Thomson Reuters who specializes in AI with an emphasis on natural-language search.

This tension finds expression in the technology. As Custis explains, “In technological terms, it’s the trade-off between precision – finding the right things – versus recall – finding all the things. If we tune an algorithm to score high in precision by delivering the ‘best’ results, there’s a trade-off in terms of some relevant results getting missed – in other words, lower recall. On the other hand, if we tune an algorithm to score high in recall by not missing relevant results, the best results will be somewhere in there, but not at the top, or maybe even buried – in other words, lower precision.”

Just as lawyers can over-delegate work to subordinates, they can also under-delegate, causing them to serve their clients less efficiently. In the context of artificial intelligence, one can imagine underutilization of AI – for example, a lawyer not using AI even though it could help that lawyer serve the client better.

In fact, given some of the psychological attributes commonly associated with lawyers – a focus on detail, a desire for control, an aversion to risk – the greater danger might very well be underutilization of, rather than overreliance upon, artificial intelligence.

“Having worked in AI for the legal profession for a long time, I know how the customer base is conservative,” says Tonya Custis of Thomson Reuters. “With Westlaw natural-language searching, lawyers will ask, ‘Why am I getting results that don’t use the specific words I searched for?’ You need to explain to the customer how the process works.”

HOW ARTIFICIAL INTELLIGENCE AND OTHER TECHNOLOGIES ARE REVOLUTIONIZING LEGAL RESEARCH

The twin challenges of not enough time and not enough confidence, as well as the tension between them, are not going away anytime soon – but they are being addressed. Powered by artificial intelligence, data analytics, and other cutting-edge technologies, Westlaw Edge confronts these challenges head-on, seeking to strike an optimal balance between saving time and increasing confidence.

Search Improvements and Suggested Answers

The heart of Westlaw Edge is the enhancement of Westlaw’s already powerful search capabilities with the latest in artificial intelligence. Westlaw has long relied upon AI in its natural-language search, but Westlaw Edge takes it to the next level. As Jon Meyer, Director of Reference Attorneys at Thomson Reuters, explains:

“Today Westlaw mostly returns documents in response to a search. With Westlaw Edge, the new search engine, WestSearch Plus, not only returns relevant documents, but it also provides highly responsive suggestions that answer thousands of types of black-letter-law questions, which can really be helpful if you’re doing research in an unfamiliar jurisdiction or topic, or if you need to get an answer quickly while on the phone.”

Legal research will always be more complex and demanding than research in the consumer space. Asking Alexa to tell you the weather is very different from asking about the constitutionality of a new statute. But these higher expectations from the consumer space are migrating into the legal space, leading lawyers to seek research tools that are faster and more intuitive.

“There’s so much new machine learning that now goes into what you put into the search bar,” says Tonya Custis. “With Westlaw Edge, we are linguistically analyzing your search. Are you looking for a group of cases about a particular subject? Are you looking for an ALR article with a question in the title? Are you looking for the answer to a specific, narrow question? We have applied artificial intelligence to help figure out user intent, so you get exactly what you are looking for.”

In other words, according to Khalid Al-Kofahi, Westlaw Edge has a greater ability to distinguish between different types of queries and react accordingly. But it doesn’t stop there, Al-Kofahi continues. Westlaw Edge helps researchers formulate the right questions, thus helping lawyers find that elusive issue or decision, the thing “on the tip of their tongue” that they can’t think of in the moment – which is just one way that Westlaw Edge saves attorneys time and increases their confidence in their research.

Of course, Westlaw Edge derives much of its power from the enhancements to the cases that have already been made – and continue to be made – by a team of attorney editors. Other legal research tools utilize artificial intelligence as well, but they are not applying AI to the information that has been enhanced, through organization and annotation, by the hundreds of highly trained and experienced attorney-editors who work for Thomson Reuters.

For example, take the predictive typeahead feature of Westlaw Edge, where a user starts typing a natural-language query and receives suggested research questions that then take the user to the exact portion of the document that address the questions. Sometimes the responses will include West headnotes, which is just one way that Westlaw Edge marries artificial and human intelligence.

“The technology wouldn’t work the same way without Westlaw’s editorial enhancements,” Leann Blanchfield explains. “If you have the technology but run it against content without enhancements, you won’t get the same results.”

The virtuous cycle runs both ways: the data enhancement helps the technology, and the technology helps the data enhancement. Artificial intelligence and other tools allow Blanchfield and her content team to move across the vast ocean of legal content more quickly, tagging cases and other authorities more accurately and consistently.

“We write 400,000 headnotes a year, and we need to be consistent and more accurate,” says Blanchfield. “If you seek consistency, think about what technology can serve up – it can identify the key numbers and headnotes that are most likely to apply to a particular case, to make sure you don’t miss anything. The attorney always has to make the final call – but the AI makes the attorney more accurate, just as the attorney makes the AI more accurate.”

 

Litigation Analytics

Using artificial intelligence to comb through a vast amount of historical data, Litigation Analytics gives attorneys a big-picture sense of how a judge might rule on a particular claim, how long such a ruling might take, and a whole host of other factors that can help lawyers in developing a litigation strategy and advising clients about how to proceed. The tool uses graphics and data visualization to present this information to attorneys in an aesthetically appealing, easy-to-understand format.

“Lawyers often formulate their litigation strategy based on their own experiences and insights,” explains Khalid Al-Kofahi. “That’s fine, but we believe we can do better. Legal Analytics is about empowering lawyers with data in addition to personal insights. It is about providing lawyers with statistics relating to the likelihood of a court or judge granting a motion, denying a motion, ruling on a motion during a specific period – you name it.”

Assessing the likelihood of a certain ruling was challenging in the past because of the sheer amount of data to be analyzed, according to Barb McGivern, Vice President of Strategy and Marketing at Thomson Reuters:

Imagine many rooms, filled from floor to ceiling with pieces of paper: court dockets. You want to know how the judge will rule on your motion, and somewhere in those vast rooms is your answer. That was the problem of Big Data in law. But now, using the latest technologies, Legal Analytics can bring those insights to the surface and present them clearly, with visualization and other tools.

“I’m so excited to talk to customers about Litigation Analytics,” says Jon Meyer. “Lawyers today use anecdotal evidence to try and figure out how successful a particular argument before a particular court or judge might be – but with Litigation Analytics, we now have the data to make informed decisions.”

Using Litigation Analytics, lawyers can quickly understand how judges might rule with a much higher degree of confidence – based on exhaustive, empirical information, not guesswork or gut instinct.

KeyCite Overruling Risk

Another area of anxiety for attorneys: whether they are relying on good law. For many years, KeyCite, Westlaw’s citation service, has flagged potential problems – literally. A red flag in KeyCite indicates that the case in question is no longer good law (on at least one point of law), and a yellow flag in KeyCite means that the case has some negative history, such as criticism from a different court.

KeyCite Overruling Risk, another feature added in Westlaw Edge, introduces an orange warning icon: an indicator that a case may be implicitly overruled, based on its reliance on an overruled or otherwise invalid prior decision. This feature allows attorneys to check the status of a case in ways that were previously not possible.

In the past, as explained by Mike Dahn, Senior Vice President of Westlaw Product Management, a status flag required an explicit citation to the flagged case in a subsequent authority. This was fine for capturing cases that got directly reversed or expressly overruled. But it failed to capture the many instances in which subsequent developments in the law turned that case into bad law.

Now, thanks to KeyCite Overruling Risk, attorneys will be alerted to situations in which a case might no longer be good law, based on new developments in the law. Lawyers will be directed to the authorities calling their case into question, allowing them to determine whether those authorities are on point – and dramatically reducing the risk of citing bad law.

How did KeyCite Overruling Risk come together? It required state-of-the-art AI technology, including advanced natural-language processing as well as traditional, supervised machine learning. KeyCite Overruling Risk was repeatedly refined during the development process, based on feedback from both Thomson Reuters attorney editors and from clients.

Insight Attorneys

Westlaw Edge is more than just a new technological platform. Westlaw Edge customers will enjoy access to “Insight Attorneys,” a special group of Westlaw research attorneys who have completed a carefully designed curriculum focusing on artificial intelligence, data analytics, and their application to legal research. The Insight Attorneys will be able to help customers craft better legal research questions and implement the new, powerful tools on Westlaw Edge for a faster, more comprehensive result.

The initial group of Insight Attorneys amounts to approximately 15 to 20 percent of Westlaw’s reference attorneys, according to Jon Meyer. But that number will grow, as more research attorneys express interest in taking the training program and as more customers make the move to Westlaw Edge.

And customers will be making the move to Westlaw Edge, as its status as a game changer becomes increasingly apparent.

“Westlaw Edge takes risk off the attorney’s plate,” says Barb McGivern. “They’ll start their research further down the line, and they’ll have more confidence in their results. Using Westlaw Edge is like having a research assistant who helps you do your best work faster.”

Despite the complexity of the technology that went into it, the appeal of Westlaw Edge is actually quite simple, according to Leann Blanchfield: “Westlaw Edge brings together the expertise and knowledge of attorney-editors with state-of-the-art technology, to help you find answers faster.”

PROFILE IN INNOVATION: THE DEVELOPMENT OF WESTLAW EDGE

As one might expect from a product of this ambition and scope, Westlaw Edge emerged out of a lengthy and exhaustive development process, involving many different teams within Thomson Reuters. The process also included external parties, including advisory boards of lawyers and law librarians, as well as design partners – lawyers at law firms who validated the initial concept for Westlaw Edge and then provided ongoing feedback to help make it into the best product it could be.

Like all Thomson Reuters products, Westlaw Edge has been designed by and for attorneys. As Barb McGivern explains:


How long was Westlaw Edge under development? It’s hard to pinpoint with precision, since it emerged out of the constant, ongoing dialogue between Thomson Reuters and its customers.

“The initial thinking for Westlaw Edge goes back several years,” Barb McGivern explains. “The concept for it was very customer-driven.”

“We worked closely with many different teams throughout the process,” Tonya Custis notes. “It was iterative and collaborative, a dialogue between the technology and the people. My team consists of researchers – we’re not lawyers, so it’s important that we get feedback from attorney-editors to make sure what we’re doing is accurate.”

We need attorneys to tell us the subtleties. When is it okay to recommend unpublished cases? When is it okay to suggest something from a different area of law – is it a useful analogy, or a waste of time? There isn’t just one easy, right answer. These are judgment calls that need to be made by lawyers, by humans – who are a crucial part of the development process at Thomson Reuters.

Although Westlaw Edge is now available in the market, the development process continues.

“I don’t think of any of our solutions as ‘done,’” says Khalid Al-Kofahi. “Westlaw Edge has now been released, but we will continue working on it, extending features and adding capabilities in response to client feedback.”

In other words, as customers start using Westlaw Edge, they will identify opportunities to make it even better. That will renew the dialogue between Thomson Reuters and its clients, starting the cycle of challenge and innovation anew.

At Thomson Reuters, innovation and technology are highly valued – but not in a vacuum. “It’s not technology for the sake of technology,” says Barb McGivern.

Instead, innovation and technology must be deployed to serve the client, who always comes first. “It might be more fun to use cutting-edge technology to solve a problem, but sometimes we might use older technology, depending on the issue,” explains Tonya Custis. “The goal is always solving the problem and helping the customer.”

Leave your email below and we'll send you free updates on Law2020.