On the quest to better understand the human brain,
researches are now putting forth effort to build a human-like robot that will have a brain
that can produce human-like artificial intelligence. This robot is looking to
be built off of a brain model called SPAUN- the Semantic Pointer Architecture
Unified Network.
SPAUN is located in a computer and has about 2.5 million
neurons, in comparison to the human brain which has 10 billion neurons, and
receives information through a pattern of voltage “spikes”. It has the capability
of recognizing structured patterns- a skill that monkeys don’t even have. It
has become unlike any other existing animal model. Nevertheless, there is still
plenty of room for potential. The brain model cannot function in real time and
therefore it is not yet considered to be a robot- just a simulation of actual
neurons. If the obstacle of structuring the brain so that it can increase its
reaction time and use the same amount of time human’s take to “think” about
problems can be overcome, then the advancement towards creating robots could be
furthered.
Although the brain cannot learn new information, it can
still be useful as model for brain damage. Chris Eliasmith did research observing
the effect of neurons getting destroyed at the same rate as humans as they age.
They were able to demonstrate that “it gets worse in the same sort of
proportion that you find in humans”.
The goal to create a brain model that is self-aware and can
produce artificial intelligence is still unreached. Do you think that SPAUN is
set in the right direction to achieving that goal? Is there a possibility in
the future that using animal models such as mice will be replaced with robots?
With several advancements in technology, the goal may be reached in our
lifetime…it’s just a matter of the path that needs to be taken.
It is highly unlikely that a robotic model is capable of replacing mice. The difference between organic and synthetic models is just too great. Simulating an organic brain is needlessly complex and expensive. It would be much more time and cost efficient to use real mouse subjects instead. However, SPAUN has other uses and further advancements in artificial intelligence and brain modeling is a welcome sight. Unfortunately, silicon based computer chips are reaching their limit. The number of transistors on integrated circuits will soon slow down and stop entirely. GPU makers NVIDIA and Intel both predict that Moore's law will flatten within the next couple of years. It will take radical technological advancements to move beyond silicon chips as well as to build a successful synthetic brain model. Unfortunately, such technology is in the realm of the far far future.
ReplyDelete