Stakeholder Groups in Computational Creativity Research and Practice - Computational Creativity Research: Towards Creative Machines - page 32

evaluation schemes from the IDEA descriptive model, e.g., change in well-being,
cognitive effort and emotional responses such as surprise and amusement.
The final two hypotheses we present above relate to communities of creative
people into which creative software is implanted. To address Hypothesis 7 , we will
need to implement software behaviours which can meaningfully be described as
subjective, and we plan to do so with the ANGELINA videogame generation system,
and others such as The Painting Fool automated artist. With such systems, we can
experiment to see whether members of the creative community are more impressed
by subjective software or not. Such an experiment could be simultaneously used
to address the final hypothesis, with knowledge of the computational origins of
artefacts systematically withheld in order to see whether positive or negative biases
hold in different creative communities. Similarly, experiments where participants
are told about the intellectual growth of a system could be carried out, to see if this
influences their impression of the software. An analysis of the findings from such
experiments could help pave the way for software to be full members of these kinds
of communities.
Looking at the three stakeholder groups studied here, we see some emerging
generalities. In particular, looking at behaviours where systems exhibit subjectivity
and intentionality, it seems clear that in all three groups, personality modelling in
software has the potential to increase the impression that people have of what software
does and, in turn, what it produces. This is part of a new understanding of creative
acts as being potentially interesting, even dramatic, episodes of activity which can
amuse and engage people, rather than a means to the end of producing an artefact of
value. This is in contrast with the traditional idea that the value of the output from
software can increase people's appreciation of the creativity it exhibits. While the
traditional view is often correct, it is not the only model of managing perceptions of
creativity in software.
The hypotheses presented here are only a subset of thosewhich should be proposed
and addressed in the future of Computational Creativity research. Not addressing
such issues would be a mistake, as stakeholder perception of creativity in software
will in part dictate the number of researchers and businesses coming into the field.
Done badly, handling of stakeholder perceptions could stall the forward progress
achieved towards embedding creative software in society. As a recent controversial
example, online retailer Amazon briefly sold T-shirts with slogans such as “Keep
Calm and Rape a Lot” [ 57 ]. The T-shirt company responsible posted an apology on its
website, and insisted that the offending articles were “automatically generated using
a scripted computer process running against hundreds of thousands of dictionary
words.” This may be the first example of computer generated artefacts causing such
offence and a company—while taking responsibility—blaming generative software
for poor quality artefacts, while tacitly acknowledging that the software had taken
on unsupervised creative responsibilities in their workplace.
Situations where software is employed independently for creative purposes in
commerce and elsewhere are likely to become more commonplace in the future. As a
more positive example, IBMresearchers have recently undertaken research to explore
the commercial potential of Computational Creativity [ 58 ], with particular emphasis