May 2020, ScientificAmerican.com 7SCIENCE AGENDA
OPINION AND ANALYSIS FROM
SCIENTIFIC AMERICAN’S BOARD OF EDITORSJOIN THE CONVERSATION ONLINE
Visit Scientific American on Facebook and Twitter
or send a letter to the editor: [email protected]State and local authorities from New Hampshire to San Fran-
cisco have begun banning the use of facial-recognition technol-
ogy. Their suspicion is well founded: these algorithms make lots
of mistakes, particularly when it comes to identifying women and
people of color. Even if the tech gets more accurate, facial recog-
nition will unleash an invasion of privacy that could make ano-
nymity impossible. Unfortunately, bans on its use by local govern-
ments have done little to curb adoption by businesses from start-
ups to large corporations. That expanding reach is why this
technology requires federal regulations—and it needs them now.
Automated face-recognition programs do have advantages,
such as their ability to turn a person’s unique appearance into a
biometric ID that can let phone users unlock their devices with a
glance and allow airport security to quickly confirm travelers’ iden-
tities. To train such systems, researchers feed a variety of photo-
graphs to a machine-learning algorithm, which learns the features
that are most salient to matching an image with an identity. The
more data they amass, the more reliable these programs become.
Too often, though, the algorithms are deployed prematurely. In
London, for example, police have begun using artificial-intelligence
systems to scan surveillance footage in an attempt to pick out
wanted criminals as they walk by—despite an independent reviewthat found this system labeled suspects accurately only 19 percent
of the time. An inaccurate system could falsely accuse innocent
citizens of being miscreants, earmarking law-abiding people for
tracking, harassment or arrest. This becomes a civil-rights issue
because the algorithms are more likely to misidentify people of
color. When the National Institute of Standards and Technology
reviewed nearly 200 facial-recognition systems, it found that most
of them misidentified images of black and East Asian people 10 to
100 times more often than they did those of white people. When
the programs searched for a specific face among multiple photo-
graphs, they were much more likely to pick incorrect images when
the person being tracked was a black woman.
Some companies are attempting to improve their systems by
feeding them more nonwhite and nonmale faces—but they are
not always doing it in ethical ways. Google contractors in Atlanta,
for example, have been accused of exploiting homeless black peo-
ple in the company’s quest for faces, buying their images for a
few dollars, and start-up Clearview AI broke social media net-
works’ protocols to harvest users’ images without their consent.
Such stories suggest that some companies are tackling this prob-
lem as an afterthought instead of addressing it responsibly.
Even if someone releases improved facial-recognition software
capable of high accuracy across every demographic, this technol-
ogy will still be a threat. Because algorithms can scan video foot-
age much more quickly than humans can, facial recognition
allows for constant surveillance of a population. This is already
happening in China, where the authoritarian government is using
the tech to suppress its Uighur ethnic minority and zero in on
individuals’ movements. These systems can easily be used to treat
every citizen like a criminal, which destroys individual privacy,
limits free expression and causes psychological damage.
In a democratic country such as the U.S., the government
needs to protect all its citizens against these kinds of measures.
But existing bans on the technology create an inconsistent patch-
work of regulations: some regions have no restrictions on facial
recognition, others ban police from applying it, and still others
prevent any government agencies or employees from using it.
Federal regulations are clearly needed. They should require the
hundreds of existing facial-recognition programs, many created
by private companies, to undergo independent review by a gov-
ernment task force. The tech must meet a high standard of accu-
racy and demonstrate fairness across all demographic groups, and
even if it meets those criteria, humans, not algorithms, should
check a program’s output before taking action on its recommen-
dations. Facial recognition must also be included in broader pri-
vacy regulations that limit surveillance of the general population—
because other identification tools that flag people based on their
gait or even their heartbeat pattern are already in development.
Americans have always been fiercely protective of the right to
privacy. Technologies that threaten that must be controlled.MACIEJ FROLOW
Getty Images
Get Out
of Our Faces
Facial-recognition tech is invasive,
inaccurate—and spreadingBy the Editors