Stakeholder Groups in Computational Creativity Research and Practice - Computational Creativity Research: Towards Creative Machines - page 13

experiment” situation in which knowledge about the personality of the artist and
their practice is entirely missing. The philosophical grounding of such an approach
[ 34 , 35 ] matches the motivation of several art movements [ 36 , 37 ] and many indi-
vidual artists who have expressed a desire for their work to be taken at face value
(see [ 38 ] for examples and further discussion).
We argue that in modern culture, a curious thing can happen when artists attempt
to remove all reference to themselves and their process from discussions about the
artistic (and commercial) value of their work. That is, in the absence of such infor-
mation, people may tend to fill in the gaps about personality and process, and may
do so in ways which bolster the credibility of an artist and increase the perceived
value of his/her works. Indeed, one could argue that—in the same way that artists
invite people to interpret the imagery in artworks in their own way by not prescrib-
ing what people should see/read/hear, in refusing to provide meta-level details about
personality and process, artists, writers and musicians are actually (purposefully or
not) inviting art lovers to invent interesting and engaging back-stories about who
they are and what they do.
In such a context of non-disclosure, the comparison of the situation for computer-
generated artefacts with the situation for human-produced artefacts is not particularly
favourable. The vast majority of people have little or no idea about programming
or programs, and may even harbour a desire not to find out about these things.
Thus, when invited to assess a computer generated painting or poem, say, without
background knowledge, they are denied any opportunity to invent a back-story, as
they cannot project personality traits or romantic situations onto the computer, and
cannot enter into any dialogues. More importantly, this situation can lead to people
realising how much they value the human connection, whether actual or imagined,
in such situations. We posit that there is a humanity gap that must be faced by
Computational Creativity researchers who want their software to enhance society by
being creative for artistic and utilitarian purposes.
Turing-style experiments, which epitomise the practice of non-disclosure, are
intended to reduce variables so that a scientific study of the value of computer gen-
erated artefacts can be undertaken. One could argue that these contexts are intended
to help people realise how much they value the aesthetic appeal of art, literature and
music, regardless of other factors. This may be true, but we believe that such tests
can actually help people realise how little they can relate to the computational origin
of artefacts. In [ 39 ], we raise other issues with Turing-style comparison studies: in
particular, we suggest that they encourage naïvety and pastiche generation in creative
software. As a final point, it is clear that such experimental conditions are not sus-
tainable if we are to enhance society with creative software. In the long term, biases
about machine creation need to be embraced and managed, rather than factored out
through experimental setups.
Hypothesis 3 Turing-style comparison tests serve to highlight the humanity gap,
and while they might serve short-term scientific gain, they are damaging to the long-
term goal of embedding creative software in society.