Learning to Be Human
Brett (Berkeley robot for the
elimination of tedious tasks) is a
humanoid robot that has taught itself
how to build and sort objects.physics. It is beyond surprising; it is mag-
ical.It’slikewatchingafishevolveintoa
human being in 40 seconds.
“The way the robot moves and begins to
walk–italmostlooksalive,”Isay.
Abbeel smiles. “Almost.”D
espite how it’s por-
trayed in books and movies,
artificialintelligenceisnota
synthetic brain f loating in a
case of blue liquid somewhere.
Itisanalgorithm–amathe-
matical equation that tells a
computer what functions to perform (think
ofitasacookingrecipeformachines).Al-
gorithmsaretothe21stcenturywhatcoal
wastothe19thcentury:theengineofour
economy and the fuel of our modern lives.
Without algorithms, your phone wouldn’t
work. There would be no Facebook, no
Google,noAmazon.Algorithmsschedule
flights and then fly the airplanes, and help
doctorsdiagnosediseases.“Ifeveryalgo-
rithm suddenly stopped working, it would
be the end of the world as we know it,”
writesPedroDomingosinThe Master Al-
gorithm,a popular account of machine
learning.IntheworldofAI,theHolyGrail
is to discover the single algorithm that will
allow machines to understand the world –
the digital equivalent of the Standard
Modelthatletsphysicistsexplaintheoper-
ations of the universe.Mathematical algorithms have been
around for thousands of years and are the
basis for modern computing. Data goes in,
the computer does its thing, and the algo-
rithmspitsoutaresult.What’s new is that
scientists have developed algorithms that
reverse this process, allowing computers to
write theirownalgorithms. Say you want to
fly a helicopter upside down: You write an
algorithmthatgivesthecomputer infor-
mation about the helicopter’s controls (the
input data), then you tell it how you want
to fly the helicopter, and at what angle (the
result), and then, bingo, the computer will
spit out its own algorithm that tells the he-
licopter how to do it. This process, called
machine learning, is the idea behind AI: If
a machine can teach itself how to fl y a heli-
copter upside down, it may be able to teach
itselfotherthingstoo,like how to fi nd love
on Tinder, or recognise your voice when
you speak into your iPhone, or, at the outer
reaches, design a Terminator-spewing Sky-net. “Artifi cial intelligence is the science of
making machines smart,” Demis Hassabis,
co-founder of DeepMind, has said.
We are, of course, surrounded by smart
machines already. When you use Google
Maps, algorithms plot the quickest route
and calculate tra c delays based on real-
time data and predictive analysis of tra c.
When you talk to Google Voice, the ability
to recognise your speech is based on a kind
of machine learning called neural networks
that allows computers to transform your
words into bits of sound, compare those
sounds to others, and then understand your
questions. Facebook keeps unwanted con-
tent of the site by scanning billions of pic-
tures with image-recognition programs
that spot beheading videos and dick pics.
Where is the acceleration of smart ma-
chines heading? It took life on Earth 3 bil-
lion years to emerge from the ooze and
achieve higher intelligence. By contrast,
it took the computer roughly 60 years to
evolve from a hunk of silicon into a machine
capable of driving a car across the country
or identifying a face in the crowd. With
each passing week, new breakthroughs
are announced: In January, DeepMind
revealed it has developed an algorithm
named AlphaGo that beat the European
champion of Go, an ancient Chinese board
game that is far more complex than chess.
Of course, humans had a hand in this rapid
Photograph bySpencer Lowell evolution, but it’s hard not to think we have85