Creative
Evolutionary Systems
Edited by
Peter Bentley and David Corne
2002, Academic Press, San Diego
571 pp, illus. B/w & col.
ISBN 1-55860-673-4
Reviewed by Robert Pepperell
The question
of how technology might intervene in human
creative processes, and whether indeed
it may surpass or supplant human creativity,
has been a recurring theme during the
rapid assimilation of computer systems
into the creative arena since the 1980s.
Because market economics habitually seeks
more efficiency in production, it would
seem inevitable that the costly vicissitudes
and unpredictable temperaments of the
artist, designer or composer would be
ripe for automation. Classical AI methods
having been shown as too brittle and clumsy,
researchers have increasingly turned to
natural principles of data
generation and manipulation to make progress.
The two principles that guide creative
evolution, the subject of this book,
are the same as those generally understood
to apply to natural evolution, namely
variation and selection. In the case of
creative evolution the principles are
applied deliberately with the intention
of using computers to generate output
of artistic, musical or design value,
often as a way of shortcutting or circumventing
the rather laborious and haphazard process
of human creative production. As the editors
put it:
" Our evolutionary techniques are
being used to automatically find solutions
to problems that traditionally required
creative people. Using evolution, our
computers are beginning to find inventive,
novel, surprising, and exciting solutions.
Creative evolutionary systems both help
us to be creative and give the appearance
of being creative themselves." (p.
xxvii)
Here, the editors have gathered what they
regard as a definitive collection of papers
covering the current state of research
in the field, drawing on a varied panel
of technologists, theorists and creative
practitioners. The volume runs to nearly
600 pages and features essays by some
50 contributors who, in different ways,
describe the methods by which computational
processes can be harnessed to creative
ends. Most of the contributions outline
how techniques of inducing random variation
can be combined with selection, often
through the subjective intervention of
a human agent or operators, to arrive
at some otherwise improbable outcome,
be it urban architecture, fractal-based
Iris prints, novel fighter pilot manoeuvres,
electromechanical robots, or musical compositions.
Well illustrated,
indexed and with an accompanying CD-Rom,
this would seem to be a definitive text
in this field.