With every technology break through, there are inevitably the glamour use cases that capture the public eye. Often, though, the business value of that technology is first realized in sectors that draw less media interest.
Machine learning/artificial intelligence (AI) is one such tech innovation which is capturing the headlines, in high-profile areas such as autonomous vehicles and vision systems in particular. Yet there is a wealth of R&D labs and startups across Canada that are taking AI to the next level in ways that are delivering results in unexplored sectors.
Toronto-based Blue J Legal is a prime example. The startup was launched in 2014 by University of Toronto professors Benjamin Alarie, Anthony Niblett and Albert Yoon, along with software engineer Brett Janssen. The outcome of their combined efforts is Tax Foresight, a simulation tool that uses AI to digitally re-master the case law that applies to specific legal questions.
In simple terms, Blue J Legal software finds links to relevant cases and generates an analysis for use in litigation, said CEO Alarie, who is the first to admit that law may not be at the top of anyone’s AI list. “There is art to this process, as well as data science, since law is rather ‘small n’ compared to a lot of the more conventional machine learning ‘Big Data’ applications.” The legal expertise required in this solution means that three of the company’s co-founders are law professors, while staff for the most part includes legal researchers with law degrees who are responsible for assembling the data with which to train the AI algorithms.
The Blue J Legal team has been working with Watson APIs as well as Microsoft Azure learning tools. They came up with their first simulation prototype in 2016, designed to solicit feedback from the field, and are now getting ready for commercialization in 2017.
Alarie believes law is a field that is ripe for innovation on the AI front. A major challenge in law is that it is an adversarial system that requires litigants put their best arguments forward so judges can make decisions, he explained. Each resolution helps to form the basis for common understanding of how judgments get made.
Over the years, active judgments and litigants number in the “thousands and thousands and thousands,” which means researchers are only able to work with a comparatively small portion of the whole opus. “There’s an opportunity from using AI in the law that would allow one to make use of all those judgments rather than a select few. Machine learning and AI actually relieve you of some of the research constraints by pooling all the available information and providing immediate insight into what the outcome would be if a judge were to consider a case.”
Legal simulation tools can provide insight based on all cases in a matter of minutes, whereas the traditional approach would take on average two weeks and provide insight based on only a portion of judgments researched, he added. This can translate into a number of benefits, ranging from improving speed to resolution and lowering costs to freeing up a lawyer’s time to work on more complicated issues that don’t lend themselves to the use of automated tools.
The idea for Tax Foresight came when Alarie was asked to judge an IBM-sponsored competition at the Department of Computer Science at the University of Toronto. Alarie noted: “All students pitching had different legal apps. As I sat there it became clear that tax law would be ideal for machine learning and AI. Although all legal researchers have used Big Data to analyze judicial decision making in the past, I became intrigued with the emergence of machine learning algorithms and what it would mean in terms of efficacy for this kind of research. As we learned more, we became excited about what actually can be done.”
Governments, accountants and lawyers struggle with complexities of taxes as least as much as tax payers, he added. “I realized that if we could build a product, there would be huge potential for both accountants and lawyers.”
While Blue J Legal’s current focus is on Canadian tax law, Alarie said the US and Australian markets are on the radar, not to mention other areas of law that involve litigation, such as family, immigration, insolvency, IP, contracts, and labour law. “Every area will be colonized by these sorts of tools. It’s just a matter of focusing on the biggest targets first.”
Alarie is convinced that AI and machine learning will improve the effectiveness of legal counsel in many different areas of law and increase the speed and accuracy of legal research analysis. “Simulation tools allows them to pin down the likelihood of success with particular legal questions. At the same time, assessing a new case, legal question or risk assessment can be reduced to a couple of hours.”
As for the future, Alarie foresees creation of a veritable mini-industry as the tools become more visible and better known. “There are thousands of people dedicated to the effort of systematically analyzing law using AI. There are lots of reasons to think adoption will be relatively quick. First adopters will have a competitive advantage when sitting across the table from counsel and making arguments their counterparts can’t respond to. In a short time, those others will have to keep pace. It will be a bit of an arms race for early adopters and fast followers.”
