Technologies fuelled by artificial intelligence (AI) are already finding their way into the stores where we shop online and into futuristic cars that drive themselves. And now a U of T legal services company is poised to use machine learning to revolutionize the practice of law.
Launched in 2014 by three U of T law professors and a veteran software engineer, Blue J Legal has created sophisticated AI software that provides lawyers and judges with guidance on resolving tax disputes. While a judge may use a dozen key precedents to make a ruling, Blue J’s technology sifts through hundreds of past cases, looking beyond key words for facts similar to the case in dispute. With successive refinements, Blue J’s accuracy – the number of times its conclusions align with a previous judge’s rulings – has improved from 65 per cent to up to 98 per cent.
Blue J got a leg up through a pilot program for machine-learning startups at Rotman’s Creative Destruction Lab. The intensive nine-month program pairs new ventures with successful technology entrepreneurs who provide crucial advice about achieving key objectives and connecting with investors.
The pilot was a success, and now the Creative Destruction Lab will use $1 million from the Rotman Catalyst Fund to enhance its machine-learning stream. The $45-million Rotman Catalyst Fund was created last year through a visionary $30-million seed gift from Joseph Rotman (MCom 1960) to fund bold and innovative initiatives that align with the school’s highest priorities. Further support for the machine-learning stream has come through gifts from Scotiabank and RBC.
According to Benjamin Alarie, Blue J’s CEO and a U of T law professor, the company’s technology can dramatically increase access to legal advice for consumers by lowering its cost. The software will also reduce the number of disputes that go to trial. “There are cases where litigants are wasting a lot of time and emotional energy fighting,” he says. “Our system could lead to settlements of easy cases.”
The company traces its roots to IBM’s 2014 “Watson Challenge,” a programming contest that involved students from U of T’s computer science department. Asked to sit on the contest jury, Alarie was intrigued by the students’ ideas and approached computer science about working with the Faculty of Law to develop the concepts further. This partnership became Blue J Legal.
Law and MBA student Claudia Dzierbicki helped with the legal and market research Blue J needed to make their pitch to the Creative Destruction Lab’s machine-learning stream. As Blue J commercializes its service, Dzierbicki predicts an important early market will be employment contract disputes that can “potentially cost millions of dollars in claims by employees [if the employer] gets it wrong.” The company sees other applications in labour, family and even criminal law. Machine learning systems “will run circles around human decision-making,” says Alarie.
For Alarie, Blue J’s very existence is evidence of how smart people from seemingly disparate fields came together, with the help of astute investments by visionary donors, to create an innovation that could potentially change something as tradition-bound as the practice of law. As he says, “It’s the perfect storm of how to get things done at the university.”
In 2016, the Rotman School of Management received a historic $30-million bequest from the estate of Joseph Rotman (MCom 1960). The gift has enabled the creation of the $45-million Rotman Catalyst Fund to support bold, innovative initiatives at the school.
U of T’s 196th Birthday Quiz
Test your knowledge of all things U of T in honour of the university’s 196th anniversary on March 15!
Spreading the Gospel
A Juno Award-winning teacher wants all his students to feel there is a place for them in music
Cities Are Driving Evolution
Globally crowdsourced study shows that white clovers are biologically adapting to city life, demonstrating the profound impact of urbanization
2 Responses to “ Helping Machine-Learning Startups Succeed ”
While reading this article, I began to wonder whether using AI in the law was really a good idea. If we rely too heavily on machines to examine precedents and recommend solutions, will we lose our ability to make reasoned analyses on our own? Will reliance on AI-recommended solutions slow adaptation to changing social conditions?
Decision-support models are typically validated by their ability to reproduce past decisions. They assume that the foundation (laws, regulations, societal expectations, etc.) is static. Use of the model will then repeat the same decisions and can stifle innovation and new thinking. In reality, the foundation is dynamic and new analysis may be required as we adapt to new laws or to changing societal expectations. Will AI models help us to adapt to the changing balance between public security (where the state might have access to significant amounts of personal information), for example, and the protection of personal privacy? Or will they hinder us?
If we continue to challenge, refine and review AI models and tools with the same degree of rigour as we do other theories and models, then my questions should be addressed. I applaud the team that collaborated to bring together legal, software, and business communities, facilitated by U of T, to develop a new tool. But let it be used wisely.
BSc 1989 Victoria
Benjamin Alarie, CEO of Blue J Legal, responds:
At Blue J Legal, our AI legal tools promote legal transparency and access to justice. The genius of the common law is that impartial judges decide cases on their relative merits and produce public reasons. The value of what we’re doing is that we are operationalizing the cumulative wisdom of these judges for the benefit of those who seek to comply with and follow the law. Too many cases are now brought to court that would have been settled if only the mutual distrust of the litigants had been overcome. With the benefit of our tools, increased rates of settlement will leave judges with more time to focus on the truly borderline cases. This will allow for an improved articulation of the boundaries of the law. Everyone will benefit.