EXAMPLE PAGE - SCHOOL BROCHURE - SMITH BUSINESS - Flipbook - Página 17
BRAIN TRUST
BRAIN TRUST
The end of the shopping mall?
Machine Learning Has a Flaw.
It’s Gullible.
Machine learning technologies are
poised to supercharge productivity in the
knowledge economy, transforming the
future of work. But they’re far from perfect.
ML—technology in which algorithms
“learn” from existing patterns in data to
conduct statistically driven predictions
and facilitate decisions—has been found in
multiple contexts to reveal bias. Remember
when Amazon.com came under fire for a
hiring algorithm that revealed gender and
racial bias? Such biases often result from
slanted training data or skewed algorithms.
And in other business contexts, there’s
another potential source of bias. It comes
when outside individuals stand to benefit
from the bias and strategically alter the
inputs—perhaps when applying for a job
or making an insurance claim. In other
words, it happens when they’re gaming
the ML systems.
But maybe ML can correct for such
strategic behavior.
In new, highly acclaimed research,
Maryland Smith’s Rajshree Agarwal and
Evan Starr, along with Harvard’s Prithwiraj
Choudhury, explored what firms can
do about attempts to outthink their ML,
focusing on patent applications—a context
rife with potential trickery.
Patent examiners face a challenge
of accurately determining how novel
and nonobvious a patent application is,
by sifting through the ever-expanding
amount of inventions that have come before.
14
/
SMITH BUSINESS
/
fall 2020
I
t’s been a rough year for shopping malls. Stay-at-home orders left
those wide glass doors locked for weeks and weeks. When they
finally could reopen, many people opted to stay away, not wanting
to be indoors in a crowd with no COVID-19 cure in sight.
The shopping mall might never be the same again, said Maryland
Smith’s Jie Zhang.
The coronavirus pandemic arrived at a time when many of
America’s iconic, enclosed shopping malls were already struggling,
with consumer traffic shifting to e-commerce, discount outlets and
big-box retailers.
“This is another big heavy blow to shopping malls,” said Zhang.
“But it’s also a catalyst.”
Even before stay-at-home measures became the norm across
the country, Forever 21, Papyrus and Modell’s Sporting Goods
announced plans to close all of their brick-and-mortar locations.
Candles stalwart Pier 1 Imports had entered Chapter 11
bankruptcy protection and announced it would close nearly half of its
936 stores. Macy's, Office Depot, Bose, Express, Hallmark, JCPenney,
Kmart, Sears, and Bed, Bath and Beyond were also closing stores.
Even lux retailer Bloomingdales closed one location—the one south
of Miami.
For malls, the impact of closures can be considerable. Empty mall
storefronts erode consumer traffic, reducing sales prospects for
neighboring shops, and diminishing overall commercial property
It’s challenging work. Compounding
that challenge: patent applicants are allowed
to create hyphenated words and assign
new meaning to existing words to describe
their inventions.
It’s an opportunity, the researchers
explained, for applicants to write their
applications in a strategic, ML-targeting way.
“Although it is theoretically feasible for ML
algorithms to continually learn and correct
for ways that patent applicants attempt to
manipulate the algorithm, the potential
for patent applicants to dynamically
update their writing strategies makes it
practically impossible to adversarially
train an ML algorithm to correct for this
behavior,” they said.
The researchers, with observational
and experimental studies, found that
patent language changes over time,
making it really tough for any ML tool
to operate perfectly on its own. The ML
benefitted strongly, they found, from
human collaboration.
That’s because people bring relevant
outside information to correct for
strategically altered inputs. And with
vintage-specific skills—skills and knowledge
accumulated through prior familiarity of
tasks with the technology—people are
better able to handle the complexities in
ML technology interfaces.
Together, ML and human intelligence are
a far more effective tool. /KJ/
values. One failed outlet begets another, and another. Zhang,
who studies consumer purchase behavior and retail strategies in
the digital and multichannel retail environments, said the era of
social distancing is forcing a shakeup for struggling enclosed
shopping centers.
For open-air shopping centers, the picture may be
rosier, with consumers feeling more “ease of mind” in
spaces that aren’t enclosed.
Already, open-air shopping centers had been rising
in popularity, helped by the appeal of being outdoors
and by their attractive retail and food establishments.
“For some shopping malls and struggling retailers, this is a very sad
story,” Zhang said. “But the coronavirus crisis may also serve as a kind
of shock that will force shopping malls to work harder to improve
their amenities.”
She sees the number of shopping malls across the United States
being scaled back in number, while being scaled up in luxury amenities.
That means more high-end restaurants, virtual try-on dressing
rooms and smart mirrors, and high-tech customized product offerings
that entice shoppers to return again and again.
“Those who can weather this storm and survive may have the
opportunity to thrive in the future when consumers eventually reach
a point where they are ready to indulge themselves,” Zhang said.
“That's the bright side.” /KJ/
“The coronavirus crisis may also
serve as a kind of shock that will
force shopping malls to work harder
to improve their amenities.
—JIE ZHANG
rhsmith.umd.edu
rhsmith-editor@umd.edu
”
fall 2020
/
SMITH BUSINESS
/
15