There has been a lot of media buzz recently about the potentials and limitations of Artificial Intelligence in a general domain of creativity. I have been studying aspects of philosophy, psychology, physics, logic, computer science and art (i.e. creativity) for quite a long time now. A question of Artificial Intelligence as creative agent creates a fascinating nexus of all these fields of study and will inevitably be a topic I return to many times. A recent advertisement for a public debate at Oxford University (by diverse SMEs) on the possibility and relationship of machine intelligence to creativity inspired the following questions.
• In what ways does a primary R&D commitment to commercial utility constrain the creativity of system developers to a limited sub-set of all possible solutions?
• Is the functional product of operational AI an adaptive tool agnostic of intended context and application or do the constraints and parameters of architectural teleology inevitably influence and shape longer-term developmental trajectories?
• If creativity is always to some extent an open, intuitive and recombinatory exploration of a conceptual possibility-space, does (or can) “machine intelligence” possess an aptitude to identifying when they have completed their creative process or will an authentically “free” creative system continue endlessly layering complexity into their product? (Consider Google Deep Mind’s disturbing hallucinatory dystopias of object reconstruction – does the Halting Problem apply here?)
• What are the logical, axiomatic limitations of machine intelligence in regards to creativity, intuitive leaps of insight? Are there any? How do the Incompleteness Theorems apply here?
• What does (the potential for) machine creativity tell us about our own?