Robotics Reference
In-Depth Information
Genetic programming
42
has proved to be another route from within
the field of AI to automatic invention. The founding father of genetic
programming, John Koza, was able to report, in 2000, in the very first
issue of the journal
Genetic Programming and Evolvable Machines
,onten
potential products that had been created by genetic software, in each
case starting with nothing more than the basic parameters of a prob-
lem posed by the programmers. In one instance, Koza's team asked the
program to design a device for receiving television signals, providing the
software only with a few requirements as to the approximate size and
performance of the desired device. The software then “invented” an an-
tenna which turned out to be the same as the familiar ladder-shaped
Yagi-Uda antenna, designed by two Japanese scientists in the 1920s and
still the most common type of terrestrial TV antenna to be found on
the rooftops of houses today. That Koza's software had proposed a de-
sign that was already known is of no negative consequence—in fact it
proves the power of genetic software for inventing useful and commer-
cially viable products. As of December 2004 the web site
http://www
.genetic-programming.com/
listed 36 instances where genetic program-
ming has produced a
human-competitive result
.
43
These results include
15 instances where genetic programming has created something that ei-
ther infringes or duplicates the functionality of a previously patented
twentieth-century invention, six instances where genetic programming
has done the same with respect to a twenty-first century invention, and
two instances where genetic programming has created a patentable new
invention. These results come from such fields as computational mole-
cular biology, cellular automata, sorting networks, and the synthesis of
designs for analog electrical circuits, controllers, and antennae.
These achievements testify to the ability of genetic programming to
act as a successful invention machine, and it is this capacity for inno-
vation that will, I believe, raise genetic programming to God-like status
in the eyes of the scientific community, if not the entire civilized world.
Eventually, all we will need to do in order to solve a problem is say to
our robots something akin to “A new illness has just been discovered in
42
See the section “Self-Reproducing Software and Genetic Programming” in Chapter 11.
43
A human-competitive result is defined for these purposes as a result that satisfies one or more of
eight criteria, including: (A) The result was patented as an invention in the past, is an improvement
over a patented invention, or would qualify today as a patentable new invention; or (B) The result
is equal to or better than a result that was accepted as a new scientific result at the time when it was
published in a peer-reviewed scientific journal.
