Rise of the Machines: Keep an eye on AI, say experts
Rise of the Machines: Keep an eye on AI, say experts
AFP March 12, 2016
A Google computer's stunning 3-0 victory in a
Man-vs-Machine face-off over the ultimate board game highlights the need to
keep Artificial Intelligence under human control, experts said Saturday.
The partly self-taught AlphaGo programme's defeat of Go
grandmaster Lee Se-Dol showed AI was progressing faster than widely thought,
they said -- a highly symbolic moment in humanity's quest to create smart
machines.
And while AI plays a key role in building a better, safer
world, some fear the fast pace of development could finally leave humans
outwitted by our own inventions.
AlphaGo's triumph "shows that the methods we do have
are even more powerful than we first thought," said AI expert Stuart
Russell of the University of California's Berkeley Electrical Engineering &
Computer Sciences department.
"The fact that AI methods are progressing much
faster than expected makes the question of the long-term outcome more
urgent," he told AFP by email.
"It will be necessary to develop an entirely new
discipline of research in order to ensure that increasingly powerful AI systems
remain completely under human control... there is a lot of work to do."
Until just five months ago, computer mastery of the
3,000-year-old game of Go, said to be the most complex ever invented, was
thought to be at least a decade off.
But then AlphaGo beat European Go champ Fan Hui, and its
creators decided to test the programme's real strength against Lee, one of the
game's all-time masters.
Game-playing is a crucial measure of AI progress -- it
shows that a machine can execute a certain "intellectual" task better
than humans.
- Advance for science -
A key test was when IBM's Deep Blue defeated chess
grandmaster Garry Kasparov in 1997.
The game of Go is more complex than chess, and has more
possible board configurations than there are atoms in the Universe.
Part of the reason for AlphaGo's success is that it is
partly self taught -- having played millions of games against itself after
initial programming to figure out the game and hone its tactics through trial
and error.
"It is not the beginning of the end of humanity. At
least if we decide we want to aim for safe and beneficial AI, rather than just
highly capable AI," Oxford University future technology specialist Anders
Sandberg said of Lee's drubbing.
"But there is still a lot of research that needs to
be done to get things right enough that we can trust (and take pride in!) our
AIs."
AI offers the promise of a highly-efficient world in
which robots take care of our sick, fly and drive us around safely, stock our
fridges, plan our holidays, and do hazardous jobs humans should not or will not
do.
In many ways it is already doing so.
But for some, unchecked AI development evokes apocalyptic
images in which hostile machines enslave humanity.
Physicist Stephen Hawking, among the leading voices of
caution, warning last year that smart computers may out-smart and
out-manipulate humans, one day "potentially subduing us with weapons we
cannot even understand."
AI specialist Jean-Gabriel Ganascia of the Pierre and
Marie Curie University in Paris, welcomed the match outcome as a major advance
for scientific knowledge.
"I don't see why we would speak about fears. On the
contrary, this raises hopes in many domains such as health and space
exploration," he said.
AlphaGo co-developer Demis Hassabis has mooted the use of
AlphaGo-type algorithms in tackling real-world problems -- from predicting
change impacts to complex disease analysis.
"In the end, the game is highly symbolic," said
Sandberg, adding that as in computer mastery of chess, "the dramatic
symbol quickly becomes commonplace".
"The AI that changes the world is not even
recognized as AI, just automation -- the algorithms routing Internet traffic,
shipment logistics, processing images and text, stock market trading and so
on," he said.
"The symbolic events are like peaks of waves, but it
is the underlying flood we should be watching."
mlr/cw
Comments
Post a Comment