Information Technology Reference
In-Depth Information
graduate or even a talented amateur artist. At least to start with, The Painting Fool's
art has been rather naive and of little cultural interest, but as we progress with the
project, we hope the value of the artworks it produces will increase. In another
respect, however, we have fairly high standards: to be called a painter, our software
must simulate a range of both cognitive and physical behaviours common to human
painters. Such behaviours naturally include practical aspects such as the simulation
of making paint strokes on a canvas. However, we are also intent on simulating
such cognitive behaviours as the critical appraisal of one's own work and that of
others; cultural and art-historical awareness; the ability to express ideas and moods
through scene composition, choice of art materials and painting style; and the ability
to innovate in artistic processes.
For some in the art world, there is a discernible resistance to using a computer in
art practice, and this is naturally heightened when mention is made of the software
acting as a creative collaborator or an independent artist. It is therefore an interest-
ing challenge to gain some level of acceptance for AI-based art producing software
within mainstream artistic communities. One problem has been that the majority of
artworks produced by software with some level of autonomy have limited appeal
and the pieces largely exist for decorative purposes. For instance, once any aesthetic
pleasure and possibly some awe at the power of modern computing has worn off,
it is difficult to have a conversation (in the cerebral, rather than the literal sense)
with an image of a fractal, or indeed many of the generative artworks that artists
and engineers regularly produce. Also, as a community of Computational Creativ-
ity researchers, there has been the assumption (or perhaps hope) that the artefacts
produced by our software—poems, pictures, theorems, musical compositions, and
so on—will speak for themselves. In certain creative domains, this may be the case.
For instance, it is possible that people will laugh at a funny joke regardless of how it
was conceived (with the caveat of controversial jokes: there is a big difference in our
appreciation of a racist joke told by a person of that race and of the same joke told
by a person of another race). However, in other domains, especially the visual arts,
there is a higher level of interpretation required for consumption of the artefacts.
In such domains, we have somewhat neglected the framing of the artefacts being
produced by our systems. Such framing includes providing various contexts for the
work, offering digestible descriptions of how it was produced, and making aesthetic,
utilitarian or cultural arguments about the value of the work. Only with this extra
information can we expect audiences to fully appreciate the value of the artefacts
produced autonomously by computers via more interesting, more informed, conver-
sations.
With The Painting Fool, we are building a system that aims to address the short-
comings described above. In particular, we are overcoming technical challenges to
get the software to produce more stimulating artworks which encourage viewers to
engage their mental faculties in new and interesting ways. These techniques include
new ways to construct the paintings, in terms of scene composition, choice of art
materials, painting styles, etc. In addition, they also include new ways to frame the
paintings, in terms of providing text about the artworks, putting them into context,
etc. We are pioneering Computational Creativity approaches which adhere to prin-
ciples designed to not only produce culturally valuable artefacts, but also to frame
