www.servomagazine.com/index.php?/magazine/article/november2011_ThenNow
Man vs. Machine
at a time. These neural cores have the ability to be
reconfigured when presented with a new task, just as our
brains do. This fall at a custom IC conference, two papers
are scheduled to be presented outlining how these chips
have learned tasks such as playing the game of Pong,
navigating a car on a clear pathway, and image recognition.
These neurosynaptic chips/neural cores are the earliest
building blocks of what IBM hopes to develop into a more
complete system — a cognitive computer. News from IBM’s
Cognitive Computing Group started making headlines in
2007 when researchers first simulated a mouse brain, then
stepped up to a rat brain, then to a cat brain in 2008, and
finally to a monkey brain. They soon realized that the
power required by a supercomputer similar to Watson (to
simulate human-scale reasoning) would incinerate itself if
placed in the small scale of a human brain. The hundreds
of kilowatts had to be reduced by a factor of ten thousand
or more, down to our brain’s power consumption of 10
watts or so. This new model that researchers were seeking
would use these low power building blocks of the neural
cores for the cognitive computer. The scientists involved
freely admit that much physical and chemical knowledge
has been learned about the human brain but exactly how
it works to produce intelligence is the still-to-be-found
“Holy Grail.”
Artificial intelligence has long been one of the most
interesting topics for robot experimenters, though some
really prefer the term ‘synthetic intelligence,’ as we are well
aware that our computer approach is not the way that our
brains process information. However, these neural cores
meld the central processing and random access memory
(RAM) sections so close together on the silicon chips, that it
is much the same as our brain’s neuron and synapses. The
resulting power requirements are minimal, and the reconfiguring ability coupled with a cool-running chip base is
the closest that we’ve come to a non-biological brain. The
software used doesn’t require the traditional programming
of logic, rules, and step-wise sequences. It uses the
interconnection of the neural cores so that the cognitive
computer learns as do we. Learning is its programming.
This has been a giant step for AI, but we humans are still
on the top of the pyramid of intelligence. We can gaze
down at Watson, the neural chip computers, and all of the
other robots, but we must realize that intelligence equality
will so be upon us.
Early Ideas Expressing
Machine Excellence
The earliest robot stories and movies always presented
a bit of distrust of robots living among humans. Robots
were always considered way too powerful to work along
side humans, and many were banished to other planets as
a safety precaution. These classic stories presented robots
as potential enemies of Mankind, often about to take over
Earth. Even later movies such as iRobot and the Terminator
series showed robots as threats to humans. Isaac Asimov’s
FIGURE 3. IBM's Watson on Jeopardy.
stories of the ‘40s and ‘50s softened our perception of
robots with the Three Laws of Robotics, but that didn’t stop
his ‘iRobot’ series of short stories from being twisted a bit
in the Will Smith movie of the same name when US Robots’
latest production run of robots seemed to go a bit crazy (to
say the least). In the mid 1960’s film, 2001 — A Space
Odyssey, surviving astronaut Dave Bowman is shown in
Figure 4 lobotomizing the spacecraft’s computer, HAL
9000. It seems HAL decided that he could run the mission
better than the five original crewmembers so he killed off
four of them, and then tried to kill Bowman. Dave proved
that man was more adept at running a spaceship than a
machine, because he could disconnect the malfunctioning
computer’s memory.
It took businesses many decades to finally trust and
accept computers. Robots followed the explosive growth of
computers, and it’s still taking some people a while to
completely accept robots in industry, as well as in the
home. Because of these misconceptions, many people
continue to feel a bit of uneasiness with the prospect of a
robot in their homes. However, I really doubt they think
that their robot vacuum cleaner will attack while they sleep.
FIGURE 4. The HAL 9000 computer in 2001 — A Space
Odessey is about to be disabled.
SERVO 11.2011 77