Information Technology Reference
In-Depth Information
Answers to these questions have many complex implications, leading to a myr-
iad of further questions, several of which arise in the dialogue that follows. When
reading this dialogue it is important to keep in mind that the
role
evaluation plays
determines the
kind
of evaluation required. Evaluation of a work as it proceeds leads
to changes in that work (and potentially to future works), creating a feedback loop
between action, intent, material (physical, musical, virtual) and decision.
Perhaps the most basic evaluation of a work as it proceeds is to know when it is
finished. Knowing when a work is “done” arises for artists in almost any medium,
working alone or in collaboration. At the opposite end—when beginning a work of
art—initial ideas, conditions, moods and decisions can have major impacts on what
follows. Cezanne reportedly threw away paintings once an “incorrect” brushstroke
was made. Improvising musicians do not have such luxuries: a pianist like Keith
Jarrett is acutely aware that what he first plays will shape the rest of the performance.
Computers have the ability to “undo”, backtrack, and trial many possible combi-
nations very quickly. But knowing what to undo, how and when to backtrack, and
which paths to pursue or abandon requires evaluation
appropriate to the task
if it
is to be successful. Evaluation of a work as it proceeds is generally concerned with
decision making and prediction, e.g. what are the implications of making this mark,
playing this sequence of sounds, or using media in a specific way? Accomplished
artists have a seemingly innate intuition about creative decision-making and its im-
pact, developed and fine-tuned over many years of practice. But can such decisions
alone lead to the transformational creativity (Boden
1991
) we see in the best human
artists?
Evaluation of a finished artwork as an “art object” presents a different set of
criteria. This may include examination of the emotional response of people experi-
encing the work under consideration, or an evaluation based on (for example) some
aesthetic principles. In this topic we are inevitably interested in what aspects of eval-
uation might be captured in a computational system. One possibility is to employ
machine learning techniques where the system is trained on existing art works in or-
der to learn any underlying aesthetic criteria. Certainly, this forms the basis of much
current research.
It is also crucial to understand what we are evaluating for:
quality
(artistic,
conceptual, aesthetic),
value
(monetary, cultural, critical, emotional), or something
else? An important distinction can be made between the evaluation of creativity
(appropriate or valuable novelty) and, for example, aesthetics. Something that is
aesthetically pleasing may not necessarily be creative (as evidenced by looking at
any collection of picturesque wall calendars, for example).
Human evaluation of artistic works typically extends well beyond the artefact
itself, encompassing implicit knowledge and cultural norms, such as the inten-
tion of the artist who created it, the situation and conditions—social, political, and
cultural—under which it is made and presented, the observer's knowledge and ex-
perience of similar works, and the dominant social values and norms of the day.
2
2
For important considerations of these issues, we refer the reader to the contributions in Part III.
