Page 38 - Robots Androids and Animatrons : 12 Incredible Projects You Can Build

Is consciousness life?
This raises a few questions: “Is intelligence conscious?” “Is con-
sciousness life?” It seems safe to say that intelligence has to reach
a certain level or critical mass before consciousness is achieved. In
any case, artificial neural networks can and will develop con-
sciousness. Whether the time span is 10 years or a 1000 years from
now makes no difference; 1000 years is less than a blink of the eye
in the evolutionary time line. (Of course, I am hoping for a 10-year
cycle so I can see a competent AI machine in my lifetime.) At the
point where an artificial neural network becomes conscious and
self-aware, should we then consider it to be alive?

Artificial life

Artificial life (AL) splinters into three ongoing research themes:
self-powered neural robots, nanorobotics (may be self-replicating),
and programs (software). The most evolved types of artificial life
on Earth today are programs. No one has created a self-replicating
robot, and nanobots are still years away from implementation.
Therefore let’s discuss AL programs for the time being.

In AL programs, life exists only as electric impulses that make 17
up the running program inside the computer’s memory. Com-
puter scientists have created diverse groups of AL programs that
mimic many biological functions (survival, birth, death, growth,
movement, feeding, sex) of life. Some programs are called cellular
automations; others are called genetic algorithms.

Cellular automation (CA) programs have been used to accurately
model biological organisms and study the spread of communicable
diseases like AIDs in the human population. These programs have
also been used to study evolution, ant colonies, bee colonies, and
a host of other chaos-driven statistics. Chaos algorithms are added
into the programs to generate randomness. One interesting appli-
cation of CA programs is to optimize neural networks running in
host computers. It is hoped that these CA programs will one day
create and wire large neural network systems in supercomputers.

Genetic algorithms (GAs) evolve in a Darwinian fashion—survival
of the fittest. Two compatible GA programs can meet in the com-
puter’s running memory, mate, and mix their binary code to pro-
duce offspring. If the offspring GA program is as healthy or has
greater health than its parents, it will likely survive.

Team LRN Artificial life and artificial intelligence