Artificial Intelligence (AI) Industries Legal Technologies

How AI is Changing Legal Research

How AI is Changing Legal Research 23/11/2017
How AI is Changing Legal Research

Photo Credit: pinkomelet/123RF

Of all the tasks that lawyers perform on a daily basis, legal research ranks on the excitement score somewhere between drafting a fee agreement and reviewing the work of a mediocre legal assistant. In fact, many lawyers dread legal research intensely and try to avoid it or pass it off to others as much as possible. Not only is it a relatively mundane task as far as legal work goes, but it is often tedious and time-consuming. The standard ways of conducting legal research can also make it difficult for a lawyer to find what they are looking for, especially if they have only a nebulous idea of what they are looking for to begin with. Luckily, artificial intelligence (AI) is stepping in to save the day. The AI revolution is upon us, and companies both large and small are developing a myriad of programs designed to overhaul the way lawyers find relevant results to their legal questions.

Comparing Apples to Oranges

Hots Viktoriia Hots Viktoriia/123RF

Have you ever wondered if the search engine or database used for legal research affects the results you’re given? Professor Susan Nevelow Mart, associate professor and director of the Colorado Law School law library, has wondered exactly this. Her curiosity led her to conduct a study on six of the leading legal research providers: Casetext, Fastcase, Ravel, Google Scholar, Lexis, and Westlaw. She published a report of her findings in a draft paper entitled “The Algorithm as a Human Artifact: Implications for Legal {Re} Search.” As it turns out, the legal research program utilized in a search makes a very significant difference to the research results obtained by the user.

The impetus to research this issue came from an email exchange between Professor Mart and Senior Vice President for Westlaw Product Management, Mike Dahn, over at Thomson Reuters. Dahn noted in one of his emails that “all of our algorithms are created by humans.” This led Professor Mart to speculate that the choices made by humans in designing these algorithms could have a notable and noticeable difference in what search results are churned out from different programs, even if users were to enter the same natural language search terms at the outset. It seemed likely that the biases of humans would translate into biases within the code. As it turns out, she was right.

Remarkably, Professor Mart found that very little overlap existed within the top 10 results of the different programs, even when identical search terms were entered. Even though each hypothetical user was looking for the same thing, they were presented with widely varying results. In addition, about 40 percent of cases returned in the results were unique to one database and only 7 percent of cases appeared in the search results for all six databases. In short, while identical search terms were used across the board, the results were a comparison of apples to oranges in terms of what the user got back from the programs. This is an issue that AI seeks to help us solve.

Utilizing AI to Combat Biases

Marina Putilova/123RF

The legal industry has high hopes for the ways in which AI will transform legal research. Already, the top two providers of legal research, Lexis, and Thomson Reuters, have rolled out AI-like programs focused on natural language processing, the past behavior of users, and relevance calculations far beyond the human ability to calculate. The goals are to produce search results that are more uniform in nature and that anticipate what the user is actually looking for even if they have done a poor or inaccurate job of entering their query terms. AI programs such as Lexis Answers and Westlaw Answers use natural language processing and AI functions to identify key phrases from a user’s query and provide ultra-responsive and relevant results.

Right now, Lexis Answers and Westlaw Answer offer limited numbers of available search queries. For example, Lexis Answers is available only for research into legal definitions, burdens of proof, legal doctrines, elements of a claim, and standards of review. Both Lexis and Westlaw are investing into their AI programs in order to improve the current offerings and add more categories for users. The AI functions also analyze user interactions, not just of one user in particular, but of all users, in order to determine what future best answers should be. This decreases the impact that any particular user’s bias has on search results, and the self-learning and self-improvement capabilities of AI mean that even a coder’s natural biases are downplayed over time as the program adapts itself to real queries and selected results.

Other Players are Appearing

rawpixel/123RF

Westlaw and Lexis are the biggest names in the legal research world, but they aren’t the only ones. Other industry players are appearing in the form of startups designed to bring AI and machine-learning to the forefront of legal research. And these companies are appearing on a global scale, not just in the U.S. Just one example of such a company is Mitra.ai in India, which uses a proprietary AI algorithm to help users narrow down search results to just those with the most relevant context. Toronto-based company ROSS Intelligence just secured $8.7 million in Series A funding earlier this year, and will use that to expand its reach on the legal research market. The ROSS platform uses AI to analyze legal issues and make connections that human users are likely to miss, and is already being used by some of the largest law firms in Canada and the U.S.

What the Future Has in Store

Lane Erickson/123RF

It’s often been said that the legal industry, as a whole, has been one of the most resistant to new technology. The time has come, however, when lawyers and law firms can no longer cover their eyes and shut their ears to the advancement of tech into their bubble. For one thing, their clients now expect that technology will be used routinely to reduce costs and provide more efficient legal results. And secondly, younger associates and partners who have grown up with technology are slowly replacing older lawyers who have perhaps been resistant to learning the necessary processes and programs. With legal research factoring so heavily into the average lawyer’s time commitment to a case or client matter, the utilization of AI programs is exactly what is needed to bring the industry firmly into a partnership with technology that has been a long time coming. Expect to see lawyers and firms continue to embrace AI-led research, and tech companies continue to find new and better ways to integrate AI into those firms.

Comments