Information Technology Reference
In-Depth Information
Another reason we believe we should not hide the fact that the artefacts are gen-
erated by a computer is because this kind of deception can set the computer up for
a fall. For instance, imagine a Turing-tester saying: “And so, I can now reveal that
these are the paintings produced by a recent art school graduate, and these are the
paintings produced by...a convicted murderer”. While this example may be a lit-
tle crass, it makes the point: by stating that the aim is to produce artefacts which
look like they might have been created by a person, it explicitly lowers the value of
the artefacts produced by computer. By using Turing-style tests, we are seemingly
admitting that pastiche is all that we aim for. At best, this shows that we don't under-
stand one of the fundamental purposes of creative endeavours, which is to produce
something interesting which no one has produced before. In many domains, there is
no right or wrong, there is only subjective impression, public opinion and the val-
ues of influential people in that domain. As there is no reason why we can't change
public opinion, there is no reason why we should compare our computer generated
artefacts to those produced by people. We can change the mind of the beholder to
more appreciate the value of the artefacts produced by our software, and in trying to
do so, we can learn a lot about the general perception of creativity in society.
Taking all the above arguments into consideration, we advocate non-blind com-
parison tests of human and computer art, where full disclosure of the processes
behind the production of each piece is given. It is not imperative that the software
generated artefacts look like they could be physically human-produced, but it might
help people to appreciate them. In such non-blind tests, if art lovers choose to buy
computer generated art as much as human art, because the pieces they buy stimu-
late their mind as well as their eye, we can claim real progress in Computational
Creativity.
1.3.7 Good Art Changes Your Mind
It is perhaps not useful to delve here into the debate about what is and what isn't art.
However, it is difficult to argue against the fact that some of the best scientific dis-
coveries force us to think more about the Universe we inhabit, and some of the best
works of art, music, and literature were explicitly designed to make their audience
engage their brains more than usual. Sometimes, the artworks are designed to make
most people engage their brains in roughly the same way, other times the artworks
are meant to be interpreted in many different ways. Sometimes, the purpose is to
engage people on a cognitive level, other times the purpose is to engage them on an
emotional level. Given this, our software should produce artefacts with the explicit
purpose of making the human audience think more. This can be achieved in a num-
ber of ways (disguise, commentary, narrative, abstraction, juxtaposition, etc.), and
some of these are easier to achieve than others.
More than any other aspect of Computational Creativity research, this sets us
apart from researchers in other areas of AI. In these other areas, the point of the
exercise is to write software to think for us. In Computational Creativity research,
