Information Technology Reference
In-Depth Information
his criteria as a toolkit out of which developers could choose to suit particular pur-
poses. Implementations of such a selection process, if included in a particular system,
would correspond to instantiations of the processes 3, 4 and 8 we have discussed.
Although Ritchie's criteria have been applied in practice, the instances known to
me [ 5 , 6 ] involve their application as a suite of criteria for the evaluation of given
systems, rather than as criteria for selecting candidates among the results. To date, I
am aware of no effort to select particular criteria to achieve specific goals. A system
capable of doing this on its own would be a significant advance.
Process 8 of establishing selection criteria based on the novelty of the results have
also been addressed for poems [ 8 ] and for narratives [ 16 ]. The need for a component
to model human cognition during the assessment phase of computational creativity
has been defended in [ 19 ].
Process 7 of reflecting on the accumulated results of the various other processes
described with a view to revising the processes involving design decisions (1, 2, 3,
4, 8) has to my knowledge not been addressed in any existing system, though the
need for iteration involving reflection over accumulated results is an integral part
of the ICTIVS model of narrative generation [ 11 ]. In many ways, the need to eval-
uate and reconsider not just the artefacts that result from a given system but also
the processes and the evaluation functions employed within the system was origi-
nally suggested by Wiggins [ 21 ] as part of his framework for understanding creative
systems. This procedure of refining the processes and functions he referred to as
“creativity at the meta-level”, and he described it as a possible way of considering
what Boden [ 1 ]termed transformational creativity within the same framework as
exploratory creativity . The concept of framing [ 3 ] included in the FACE descriptive
model addresses a static view of this reflective procedure as perceived by an external
viewer if a computational creative system is provided with the means for explaining
its actions. However, the concept of framing as described is perfectly compatible with
such explanations being hard-wired in by the programmer, either as justifications of
the procedures actually implemented—as done for instance in [ 4 ]—or even as flights
of fantasy unrelated to them. The reflective process as described here would require
the system to actually take decisions with respect to the procedure or criteria to be
employed. The framing could then be produced by providing articulate justifications
of these decisions. Proper framing would then evolve beyond a template-based expla-
nation of hard-wired constructive procedures, to become a transcription of complex
decision processes actually carried out dynamically by the machine.
19.3 Looking into the Future
When asked to provide a personal perspective on the future of Computational Cre-
ativity I had to think carefully about what I could do, but most of all, about what I
should not do. Nobody can foretell the future. So from the start I was determined not
to include wild guesses about what might be the best-performing technologies in CC
in the future, or what the killer applications will be 10 years from now. All efforts
Search WWH ::




Custom Search