Home' Technology Review : May June 2013 Contents 35
MIT TECHNOLOGY REVIEW
vol.116 | no.3
The movie 2001: A Space Odyssey
introduces a popular notion of AI
through the computer HAL.
Sir James Lighthill reports to
British officials on a “pronounced
feeling of disappointment” in AI’s
accomplishments. The report supports
declines in government funding during
the “A I winter” of the 1970s and ’80s.
IBM’s Deep Blue supercomputer
defeats world chess champion
Garry Kasparov in a six-game match.
Kasparov had beaten IBM computers
in two previous matches.
IBM’s Watson wins Jeopardy!,
defeating two of the game show’s most
successful contestants of all time.
Just what do you think
you’re doing, Dave?
is repeated in successive layers until the
system can reliably recognize phonemes
Like cats. Last June, Google demon-
strated one of the largest neural networks
yet, with more than a billion connections.
A team led by Stanford computer science
professor Andrew Ng and Google Fellow
Jeff Dean showed the system images from
10 million randomly selected YouTube vid-
eos. One simulated neuron in the software
model fixated on images of cats. Others
focused on human faces, yellow flow-
ers, and other objects. And thanks to the
power of deep learning, the system identi-
fied these discrete objects even though no
humans had ever defined or labeled them.
What stunned some AI experts,
though, was the magnitude of improve-
ment in image recognition. The system
correctly categorized objects and themes in
the YouTube images 16 percent of the time.
That might not sound impressive, but it
was 70 percent better than previous meth-
ods. And, Dean notes, there were 22,000
categories to choose from; correctly slot-
ting objects into some of them required, for
example, distinguishing between two simi-
lar varieties of skate fish. That would have
been challenging even for most humans.
When the system was asked to sort the
images into 1,000 more general catego-
ries, the accuracy rate jumped above 50
Training the many layers of virtual neu-
rons in the experiment took 16,000 com-
puter processors—the kind of computing
infrastructure that Google has developed
for its search engine and other services.
At least 80 percent of the recent advances
in AI can be attributed to the availability
of more computer power, reckons Dileep
George, cofounder of the machine-learning
There’s more to it than the sheer size of
Google’s data centers, though. Deep learn-
ing has also benefited from the compa-
ny’s method of splitting computing tasks
among many machines so they can be done
much more quickly. That’s a technology
Dean helped develop earlier in his 14-year
career at Google. It vastly speeds up the
training of deep-learning neural networks
as well, enabling Google to run larger net-
works and feed a lot more data to them.
Already, deep learning has improved
voice search on smartphones. Until last
year, Google’s Android software used a
method that misunderstood many words.
But in preparation for a new release of
Android last July, Dean and his team helped
replace part of the speech system with one
based on deep learning. Because the multi-
ple layers of neurons allow for more precise
training on the many variants of a sound,
the system can recognize scraps of sound
more reliably, especially in noisy environ-
ments such as subway platforms. Since it’s
likelier to understand what was actually
uttered, the result it returns is likelier to
be accurate as well. Almost overnight, the
number of errors fell by up to 25 percent—
results so good that many reviewers now
deem Android’s voice search smarter than
Apple’s more famous Siri voice assistant.
For all the advances, not everyone
thinks deep learning can move artificial
intelligence toward something rivaling
human intelligence. Some critics say deep
learning and AI in general ignore too much
of the brain’s biology in favor of brute-force
One such critic is Jeff Hawkins,
founder of Palm Computing, whose lat-
est venture, Numenta, is developing a
10 breakthrough technologies
4/10/13 6:14 PM
Links Archive March April 2013 July August 2013 Navigation Previous Page Next Page