Information Technology Reference
In-Depth Information
4.4 Conclusion
As you might expect from a topic as complex as computational evaluation of art,
there is no real consensus or closure from this discussion, nor could this be realis-
tically expected. Yet it is interesting to examine the different perspectives partici-
pants consider to be useful or practical in approaching computational evaluation. As
Paul Brown's concluding remarks emphasise, unless you think there is something
fundamentally uncomputable and ineffable in what humans do, then computational
modelling of human evaluation is at least a possibility. But just because something
is possible doesn't make it easy, or even practical. It is tantalising to think that future
computational models will shed a different light on evaluation of art (and more gen-
erally on human behaviour), complementing and informing other discourses such
as critical and cultural theory, or philosophical aesthetics. However, computational
models of this kind are still very much in their infancy.
It is also interesting to consider the mirror question to the one that is the main
topic of this chapter. Namely, can art made by an individual computer program
(or social network of autonomous computer agents) ever be fully understood and
evaluated by humans? Such considerations raised in this chapter, and many others
running through the entire volume, raise many crucial questions to investigating
creativity through computing, a number of which are listed in the final Chap. 16 of
this topic.
Evaluation remains a difficult and vexed issue for understanding creativity from
a computational perspective. No doubt it is something that artists and musicians are
involved with at almost every moment of their creative practice, but so far attempts
to mimic this process in a machine fall short of what any human being can easily do.
Interestingly, the two artists with perhaps the longest experience in this field (Nake
and Cohen) see little merit in pursuing the idea of developing creative or aesthetic
measures, precisely because they have tried to use them in their own art practices
and found them to be creative dead-ends. This should at least give us cause for
reflection. While understanding exactly what evaluation is and how it is performed
by humans remains an open problem, anyone wanting to make serious inroads into
developing machine creativity cannot afford to ignore it.
Acknowledgements
We acknowledge the contribution from all the participants, including the
original Dagstuhl discussion group on this topic, which consisted of Harold Cohen, Margaret Bo-
den, David Brown, Paul Brown, Oliver Deussen and Philip Galanter. The discussion group notes
can be found at
http://drops.dagstuhl.de/opus/volltexte/2009/2212/
. The interview in this chapter
was edited by Jon McCormack.
References
Boden, M. A. (1991).
The creative mind: myths & mechanisms
. New York: Basic Books.
Boden, M., d'Inverno, M., & McCormack, J. (Eds.) (2009).
Computational creativity: an interdis-
ciplinary approach
.
Dagstuhl seminar proceedings: Vol. 09291
. LZI.
http://drops.dagstuhl.de/
