Five University of Toronto (U of T) students have created an artificial intelligence to help lawyers obliterate the lengthy legal research process.
ROSS was developed out of a competition hosted by IBM that encouraged students to put its cognitive computer, Watson, to work. The competition was open to 10 universities, including U of T. The students’ goal with ROSS was to help lawyers power through legal research at a more efficient rate.
“It was a big surprise to me to learn that lawyers didn’t actually spend most of their time in legal battles and being in the court of law,” said student and researcher Jimoh Ovbiagele. “They spend most of their time conducting legal research, and it was in the back of my head that I started thinking about what Watson could do for lawyers.”
“It’s very cutting-edge technology,” he said. “Watson is a cognitive computer—it thinks like you and I do. It isn’t something that can be programmed. We had to teach it how to read the law and answer in ordinary language.”
IBM’s Watson uses information found in regular texts to answer questions. Unlike traditional computers, it uses context, culture, and grammar.
It uses learning and rationalization to interpret data. In 2011, Watson became well known for beating former Jeopardy winners in a game of Jeopardy.
ROSS is based off of Watson’s system. Lawyers ask ROSS their legal question in plain language and ROSS will derive meaning from legal texts and cases to provide answers.
It is fed bodies of Ontario corporate law information, cases and statutes and is able to analyze new decisions that could have an effect on present cases.
Currently, ROSS has access to publicly available documents, but Ovbiagele said they are working on teaching ROSS to understand more material and more areas of the law.
Vincent Kazmierski, associate professor of Carleton University’s law department, said “as a lawyer who has worked in the field, a primary concern is always that we have missed something, so reliance on ROSS would definitely depend on how effective it is.”
“If ROSS is truly effective, it might actually reduce legal fees as it can save time and effort in legal research and finding similar cases,” he said. “If ROSS could identify the nuances between cases, that would be effective as well.”
But he said ROSS could not entirely replace lawyers.
“Since ROSS relies on a formula, you will always need humans involved to analyze and interpret research,” Kazmierski said. “Artificial intelligence based on formulas can only do so much since cases don’t always follow patterns of logic—it depends on judge’s decisions. Each case is unique and should be analyzed as such.”
Ovbiagele and the team recently returned from New York City, where they pitched ROSS to IBM for the chance to have continued access to Watson and capital money. Ovbiagele said the pitch went well and they and IBM have formed a partnership to try to make ROSS a reality.
