The term ‘computational creativity’ is being emerged as a subfield of
computer science. It is a movement to solve artistic problems through
artificial intelligence. For artists, a new tool called artificial intelligence is not
yet familiar. So we ...
The term ‘computational creativity’ is being emerged as a subfield of
computer science. It is a movement to solve artistic problems through
artificial intelligence. For artists, a new tool called artificial intelligence is not
yet familiar. So we need to explore how we can embrace this tool in art.
The purpose of this research is to provide timely and meaningful implications
for the artist and the arts community by empirically verifying it from an
academic point of view.
This paper empirically demonstrates the relationship between art and
technology through technology acceptance and art appreciation. First,
in <Study 1>, artificial intelligence acceptance in art is verified based
on innovation acceptance theories and artistic expertise. Second, in
<Study 2>, art appreciation on works generated by artificial
intelligence and created by human artists is verified through ‘Reversed
Turing Test’.
In <Study 1>, ‘artificial intelligence’ was selected as the target
technology to confirm the acceptance of innovation. Many researchers
have already reaffirmed variables such as innovativeness, social
influences, perceived innovation characteristics, and innovation
resistance, which have been proven to influence innovation acceptance.
In addition, this paper introduce a new variable called ‘expertise’ to
examine the impact on innovation acceptance. In this study, in order to
specify the influence of art expertise, the scope of research was
restrained. This study only focus on ‘art expertise’ on ‘acceptance of
artificial intelligence in art creation’.
<Study 2> statistically demonstrates the attitude of art appreciation
on works generated by artificial intelligence. To do this, Deep Dream
Generator from Google was selected as a machine painter. Three
people, who believe themselves as artists, and also were graduated
from art school, were selected as human painters. In order to evaluate
the ability of feature extraction and the quality of artistic value, AI
painter and human painters were invited to ‘Reversed Turing Test’. If
human works are evaluated significantly lower than a machine’s
works, it is regarded as ‘failure to pass the Reversed Turing test’.
<Study 1> has three major results. First, as a result of examining
the effect of innovation variables on acceptance of artificial intelligence
in arts, novelty seeking, subjective norm, social image, and relative
advantage are positively significant, and innovation resistance is
negatively significant. Self-efficacy and perceived risk were
insignificant. Second, as a result of examining the influence of
expertise on acceptance of artificial intelligence in arts, art knowledge
and art experience, which are introduced as variables of expertise in
this study, all have a significant influence on the acceptance of
innovation. Art knowledge is positively significant, and art experience
is negatively significant. Third, as a result of comparing the difference
between the art students and non art students, innovativeness between
art and non art students is not so much different. When it comes to
significantly different variables(novelty seeking, subjective norm,
innovation resistance), the innovativeness of art students was even
relatively lower than non art students. In addition, if art students have
art expertise, it has no effect on accepting artificial intelligence in arts.
On the other hand, when non art students have art expertise, it
influences acceptance of artificial intelligence in arts.
As a result of the analysis of <Study 2>, it was found that the
machine works were significantly higher than the human works in
terms of feature extraction and artistic value, and therefore the human
painters failed to pass the Reversed Turing Test.
Art students are likely to lower their score than non-arts students in
evaluating artistic value. More specifically, art students are likely to lower
the artistic value than the feature extraction, lower the machine’s works than
the human’s.
In <Study 1>, the introduction of 'artistic expertise' variables in the
acceptance of innovation acceptance makes this research distinguishable, and
it seems to be the first research to demonstrate the innovativeness of arts
students toward new technology. As a result of the analysis, it showed that
there is no significant difference in acceptance of artificial intelligence in art
creation between art and non-art students.
Proposing ‘Reversed Turing Test’ makes <Study 2> distinctive from
other studies on Turing Test. Especially, although it is obtained from
a limited experimental environment, it has been shown empirically that
human created artwork can be evaluated lower than machine created
artwork, and that one of the factors affecting aesthetic value can be
feature extraction.