FEATURE ON PAGE 48
sees is like the images it has been
trained on or a fake produced by the
generator—basically, is that three-armed person likely to be real?
Over time, the generator can
become so good at producing
images that the discriminator can’t
spot fakes. Essentially, the generator has been taught to recognize,
and then create, realistic-looking
images of pedestrians.
The technology has become
one of the most promising advances
in AI in the past decade, able to help
machines produce results that fool
even humans.
GANs have been put to use creating realistic-sounding speech and
photorealistic fake imagery. In one
compelling example, researchers
from chipmaker Nvidia primed a
GAN with celebrity photographs to
create hundreds of credible faces
of people who don’t exist. Another
research group made not-uncon-vincing fake paintings that look like
the works of van Gogh. Pushed further, GANs can reimagine images
in di;erent ways—making a sunny
road appear snowy, or turning
horses into zebras.
The results aren’t always perfect: GANs can conjure up bicycles
with two sets of handlebars, say, or
faces with eyebrows in the wrong
place. But because the images and
sounds are often startlingly realistic, some experts believe there’s a
sense in which GANs are beginning
to understand the underlying structure of the world they see and hear.
And that means AI may gain, along
with a sense of imagination, a more
independent ability to make sense
of what it sees in the world.
—Jamie Condli;e
Google’s Pixel Buds show the promise of real-time
translation, though the current hardware is clunky.
BREAKTHROUGH
Near-real-time
translation now works
for a large number of
languages and is easy
to use.
WHY IT MATTERS
In an increasingly
global world, language
is still a barrier to
communication.
KEY PLAYERS
Google
Baidu
AVAILABILITY
Now
In the cult sci-fi classic The Hitchhik-
er’s Guide to the Galaxy, you slide a
yellow Babel fish into your ear to get
translations in an instant. In the real
world, Google has come up with an
interim solution: a $159 pair of ear-
buds, called Pixel Buds. These work
with its Pixel smartphones and Google
Translate app to produce practically
real-time translation.
One person wears the earbuds,
while the other holds a phone. The
earbud wearer speaks in his or her
language—English is the default—
and the app translates the talking
and plays it aloud on the phone. The
person holding the phone responds;
this response is translated and played
through the earbuds.
Google Translate already has a
conversation feature, and its iOS and
Android apps let two users speak as
it automatically figures out what lan-
guages they’re using and then trans-
lates them. But background noise can
make it hard for the app to under-
stand what people are saying, and
also to figure out when one person
has stopped speaking and it’s time to
start translating.
Pixel Buds get around these problems because the wearer taps and
holds a finger on the right earbud
while talking. Splitting the interaction
between the phone and the earbuds
gives each person control of a microphone and helps the speakers maintain eye contact, since they’re not
trying to pass a phone back and forth.
The Pixel Buds were widely
panned for subpar design. They do
look silly, and they may not fit well in
your ears. They can also be hard to
set up with a phone.
Clunky hardware can be fixed,
though. Pixel Buds show the promise
of mutually intelligible communication
between languages in close to real
time. And no fish required.
—Rachel Metz
