On the Role of Computers in Creativity-Support Systems - Computational Creativity Research: Towards Creative Machines - page 116

subscribes to this extreme position, most accounts of creativity do acknowledge the
role of the audience [ 12 , 13 , 22 , 41 ].
In the audience-based view of creativity, it is generally accepted that in order for
an artifact to be deemed creative, it must be novel and useful . We have argued above
(Sec. 6.2.1 ) that novelty is cognitively difficult for people because we are constrained
by our previous conceptual associations. Researchers who study creativity have come
up with various techniques to overcome this difficulty. However, computers and AI
systems, which do not have any such associations, have a great advantage here, for
they can search the unchartered areas of novel concepts and conceptual combination
more systematically [ 19 , 31 ].
However, to automatically assess the usefulness of created artifacts is a different
cup of tea altogether. As the usefulness is necessarily from a human point of view,
the question becomes: Can an algorithm capture usefulness to humans? Here, we
can distinguish two different aspects of usefulness. One is aesthetics, which relates
to artistic creativity. In this regard, there has been some research to suggest that at
least some of our aesthetic values are hardwired in the structure of the brain [ 52 , 68 ].
Moreover, machine-learning techniques have been applied to learn about the cultural
preferences of an audience based on the past data. For instance, Ni et al. [ 45 ] trained
their program with the official UK top-40 singles chart over the past 50 years to learn
as to what makes a song popular. A program like this might successfully predict, for
instance, the winner of the future Eurovision competitions. However, a limitation of
these approaches is that they cannot predict drastic changes in the aesthetic values
and tastes: for example atonal music, or abstract art. Moreover, creativity is not the
same as popularity. So to be able to predict whether a song, or a topic, or a video
will become popular [ 63 ] is not the same thing as evaluating their creativity.
This problem becomes more severe when we move beyond arts, and consider
creativity in problem solving, and in science and technology. Here the usefulness of
a novel and creative idea comes down to simply whether it works. This clearly has an
objective component, for in a sense it is the reality that determines whether the idea
works or not. History of science and technology is full of many interesting and novel
ideas that did not work. Prehistory of flight [ 20 ] is a rich domain of examples where
many novel ideas that were based on numerous observations, experimentations, and
in which inventors had complete faith, did not work at all. The following examples
provide further support for this argument (see also [ 28 , 48 ]):
1. Schön [ 58 , 59 ] described the case of a product-development team in the 1940s,
which was working to develop a synthetic-fiber paintbrush that would yield
smoothly painted surfaces like the natural-fiber paintbrush did. They came up with
an innovative and successful design using the painting-as-pumping metaphor.
However, during the problem-solving phase, they also considered painting as
masking-a-surface metaphor, which was quite a novel idea, but it led to no useful
insight.
2. Yolanda Baie, a food stand operator and owner, petitioned to have the kitchen of
her house, where she prepared food sold at the food stand, qualify as a home office
for the purpose of tax deduction (Baie vs. C.I.R., 74 T.C. 105), using the argument