Anthony Niblett,
Associate Professor and Canada Research Chair in Law, Economics, &
Innovation at the University of Toronto Faculty of Law, and a co-founder of
machine learning start-up Blue J Legal,
spoke this afternoon to Etopia News,
and provided additional details about a reference to the firm in a recent Economist article entitled “If
computers wrote laws: Decisions handed down by data” in its special supplement “The World If.”
According to
the Economist article, Blue J Legal
is “a startup combining law and machine learning to provide answers about
complex areas of tax, such as how to determine if a person is an employee or
independent contractor.” According to
the Economist, this approach
represents the possible replacement of “rules and standards” as the basis of
legal reasoning by “micro-directives” derived by algorithms from existing case
law.
This idea is
introduced and elaborated upon by Professor Niblett and his co-author,
University of Chicago Law Professor Anthony Casey, in an article entitled "The Death of Rules and Standards," available online here.
According to Niblett,
the company’s operations are now in beta test mode, designed to provide a
proof-of-concept for the idea that gray areas in the law can be reduced to
black-and-white answers by applying machine learning algorithms to masses of
existing case law, in order to determine more definitively the answers to
questions of interest, in this initial case to questions about tax law.
You can watch a
video explanation of the origins of Blue J Legal from its CEO, Benjamin Alarie,
on the company’s Facebook page, here.
The legal data
being crunched by the algorithms is Canadian tax law, which is, conveniently
enough, uniform throughout Canada, which avoids complications generated by
multi-jurisdictional issues. The system’s
data set includes 600 instances of case law relevant to the question of whether
a particular person is an employee or an independent contractor. Instead of having to wait for a judge or the
IRS to rule on any particular instance, a tax professional or lawyer can use the
Blue J Legal system to input specific facts of the case to determine with 95% probability of
accuracy the right answer to this question.
Not only can it
provide a yes/no answer to the question, it can also provide the reasoning
behind the answer and links to the relevant cases involved.
The system is now
being used by accountants in the beta test to answer this and other questions. According to Niblett, three of the Big 4
accounting firms are using the system in the beta trial. You can see a client list here.
This process
can also be applied to questions of legal residency in Canada, which, he said,
revolve around the “center of vital interest test.”
The more
consistent the existing case law is, the higher the confidence level that can
be applied to the answer. According to
Niblett, Canadian tax case law is very consistent, yielding high confidence
levels for the system’s answers. Users
of the product are, he says, “happy with it.”
This innovative
start-up’s name derives from the blue that represents the University of Toronto
and the “J” that follows the name of judges in official documents in Canada.
Another field
of law that is susceptible to this kind of machine learning is anti-trust law. Even criminal law might be a possible area
subject to this kind of analysis and prediction. The company envisions an eventual expansion
to include U.S. law within its purview.
The company
co-founder emphasizes that their system is, essentially, a “legal research
tool.” According to Niblett, neither
judges nor lawyers are worried about being disintermediated or replaced by the
Blue J Legal system. “Not yet,” he adds.
