Washington Lawyer - October 2019 - 14

Earlier this year, San Francisco District Attorney George Gascón announced that
his office would use AI to scan police reports and remove all references to race,
allowing prosecutors to make race-free decisions when initially charging individuals of crimes.
"Lady Justice is depicted wearing a blindfold to signify impartiality of the law,
but it is blindingly clear that the criminal justice system remains biased when it
comes to race," said Gascón in a written statement announcing the AI deployment. "This technology will reduce the threat that implicit bias poses to the
purity of decisions."
But AI's critics have urged caution, noting that the same biases that creep
into human decision making can find their way into programs and algorithms, and the consequences could be far more widespread and long
lasting.
"When you start to automate with AI, you are outsourcing decision making
to the people who created that program, and you cannot know what
data influenced their decisions," says Alexandra Givens, an adjunct professor
of law and executive director of Georgetown University Law Center's
Institute for Technology Law & Policy. "It is one thing to have a single
biased judge, but the decisions coded into a hiring platform will apply to
every candidate who goes through that platform, making it a potentially
biased system."
The legal industry has already embraced AI-driven, productivity-focused
software and applications, harnessing the power of technology to master
e-discovery, predictive analytics, and contract review. Up to this point, AI's
central value has been its ability to automate and streamline processes that
are considered arduous and labor intensive inside law firms.
AI providers are hoping to eliminate the deeply ingrained biases and attitudes
that affect an individual's unconscious understanding and actions. While these
can be both positive and negative biases, they are generally activated involuntarily and without warning. They cannot be reduced through introspection,
which leaves researchers to elevate an algorithmic solution.

Companies such as Unilever, Accenture, and LinkedIn also use games in their
recruitment process.
"These games are not like any personality test that you've ever taken," says
Snyder. "They don't ask subjective questions. They are actual games that are
based on tests used in sociological literature for decades. Millions of people
have played these types of games, and we know as a result that they are
predictive."
One of the deciding factors for O'Melveny was that pymetrics provided evidence
the games are free of bias and will not institutionalize any historic bias. Law firm
leaders also liked the fact that pymetrics created a student report of the findings,
which can be shared with other potential employers that pymetrics serves. To
increase impartiality, pymetrics only offers O'Melveny three potential verdicts for
each game player: highly recommend, recommend, and not recommend.
Before implementing the new AI tools, O'Melveny hosted a series of calls this
past year with the law schools it usually visits during its summer recruitment
season. Recognizing the limits of campus visits, the firm also made the games
available online to first-year law students who were interested in applying to
the firm.
"One of our primary objectives in looking at these tools was to broaden our
pipeline of candidates," Snyder says. "It's very difficult under traditional law firm
recruiting models to recruit people other than those you meet on campus. We
were looking for something that would allow us to achieve a real and objective
evaluation and go beyond campus interviews."

BIAS IN ITS MANY FORMS
AI's smart algorithms are trained to make inferences and conclusions based
on massive data sets and machine learning. Like most technology, there is an
impression of objectivity and precision because many believe technology must
naturally be impartial.
That might have been true when evaluating contract language or reviewing a
trove of emails, but the sophisticated analysis required in human resources or

Whether the legal industry will fully embrace this latest incarnation of technology depends largely on its promoters and whether they can convince law
firm leaders and criminal justice officials that objective algorithms are better
at executing decisions than biased humans.

continued on page 16

A NOVEL HR SOLUTION FOR FIRMS
The first U.S. law firm to pioneer bias-free algorithms in the recruitment process
is O'Melveny & Myers LLP, which announced in November 2018 that it would
use neuroscience-infused online games to assess associate applicants for a
range of cognitive, social, and emotional traits.
"A number of corporations have used these types of games, but we are the first
law firm to use them," says Darin Snyder, O'Melveny's diversity and inclusion
partner and member of the firm's executive committee. "The legal profession
is sometimes slow to innovate. In this context, we're happy to be that brave soul
that goes first."
The law firm hired the New York-based company pymetrics to design games to
meet O'Melveny's specific requirements. First, the firm had its high-performing
associates play the games to generate the training data necessary to "educate"
the algorithm. AI measured the O'Melveny team's levels of effort, attention,
planning, memory, and flexibility. Using these results, and removing any gender,
race, or ethnic identifiers, pymetrics built a success profile for the games.
When law school students play the games this year in hopes of securing
an interview with O'Melveny, the algorithm will be able to cull the best candidates. The focus is on prospective success and not on pedigree, says Snyder.

WASHINGTON LAWYER

OCTOBER 2019

It's very difficult under traditional law firm
recruiting models to recruit people other than those
you meet on campus. We were looking for something
that would allow us to achieve a real and objective
evaluation and go beyond campus interviews.
DARIN SNYDER
O'Melveny & Myers LLP

Washington Lawyer - October 2019

Table of Contents for the Digital Edition of Washington Lawyer - October 2019