Ross Ashby long ago pointed out that no system (neither computer nor organism) can produce anything new unless the system contains some source of the random. In the computer, this will be a random-number generator which will ensure that the “seeking,” trial-and-error moves of the machine will ultimately cover all the possibilities of the set to be explored.
In other words, all innovative, creative systems are (…) divergent; conversely, sequences of events that are predictable are, ipso facto, convergent.
— Bateson, Gregory. Mind and Nature. A Necessary Unity. New York: E. P. Dutton, 1979.
Each subsystem has two components (as is implied by the word stochastic): a random component and a process of selection working on the products of the random component.
— Bateson, Gregory. Mind and Nature. A Necessary Unity. New York: E. P. Dutton, 1979.
In that stochastic system to which Darwinians have paid most attention, the random component is genetic change, either by mutation or by the reshuffling of genes among members of a population. I assume mutation to be nonresponsive to environmental demand or to internal stress of the organism. I assume, however, that the machinery of selection which acts on the randomly varying organisms will include both each creature’s internal stress and, later, the environmental circumstances to which the creature is subjected.
— Bateson, Gregory. Mind and Nature. A Necessary Unity. New York: E. P. Dutton, 1979.
The explicit argument against the creation of long texts by completely random processes dates back more than two thousand years, to Cicero. In his De natura deorum (“On the Nature of the Gods”), Balbus the Stoic presents the following argument against the atomists (such as Democritus), who have argued that the order of nature arose out of the random collision of atoms: “I can’t but marvel that there could be anyone who can persuade themselves that solid atoms moving under the force of gravity could construct this elaborate and beautiful world out of their chance collisions. If they believe this could have happened, then I don’t understand why they shouldn’t also think that if innumerable copies of the twenty-one letters of the alphabet, made of gold or what have you, were shaken together and thrown out on the ground they could spell out the whole text of the Annals of Ennius. I doubt whether chance would succeed in spelling out a single verse!
— Lloyd, Seth. Programming the Universe. A Quantum Computer Scientist Takes on the Cosmos. London: Jonathan Cape, 2006.
Professor Max Bense, whose theory of generative aesthetics has been based on computing procedures, has pointed out that randomness involved in computer graphics replaces that aspect in art which is described as intuitive. Thus the randomizing procedures in computer technology are analogous to an artist’s intuition. This theory is questionable but it shows that attempts are being made to find equivalence between human activities in the sphere of creativity and the realization of those activities with the aid of a cybernetic device.
— Reichardt, Jasia. The Computer in Art. London: Studio Vista, 1971.
When it comes to the construction of autonomous machines that exhibit such a prized human ability our engineering skills arguably fall short. The potential of the computer as a dynamical system generating patterns that we might easily associate with art may only be realised if we know how to engineer collections of instructions that allow it to exhibit creative ability. The characteristics required of it might be summarised as follows. The system must exhibit
— Coherence and unity, maintain its identity over time, despite perturbation;
— Multi-scaled temporal complexity, demonstrate complex dynamics over fine and coarse timescales;
— Autonomous production of novelty, explore large design spaces independently of human input;
— Responsiveness to perturbation, permit external events to deeply influence its behaviour.
An artist might also prefer that the system exhibit
— Susceptibility to external control, permit external (artist-laid) constraints on its behaviour. [all are further developed in the following pages]
— Dorin, Alan. “A Survey of Virtual Ecosystems in Generative Electronic Art.” The Art of Artificial Evolution. A Handbook on Evolutionary Art and Music. Eds. Romero, Juan and Penousal Machado. Berlin: Springer, 2008. 289-309
As neither AARON nor any known contemporary system has reached such a level of autonomy, generative works will likely continue to be seen as human works. If the role of the system exceeds that of a conventional tool, these works might be seen more as human-machine collaborations; collaboration, as used here, does not require machine autonomy.
— Ariza, Christopher. “The Interrogator as Critic: The Turing Test and the Evaluation of Generative Music Systems.” Computer Music Journal 33.2 (2009): 48-70.
Critical common sense would find the idea of an alien, machinic intelligence not only rebarbative but contradictory. Because humans program machines, machines must in principle be under the control of humans. The tacit assumption here is that it is impossible to make something autonomous. To think otherwise would be fetishism or reification and, in the case of computing, to subscribe to the dehumanizing effects of instrumental rationality.
— Goffey, Andrew. “Intelligence.” Software Studies: A Lexicon. Ed. Fuller, Matthew. Leonardo. Cambridge, Massachusetts: The MIT Press, 2008. 132-42.
Like those failed media renaissances before this one, we remain one step behind the capability actually being offered us. Only an elite—sometimes a new elite, but an elite nonetheless—gain the ability to fully exploit the new medium on offer. The rest learn to be satisfied with gaining the ability offered by the last new medium. The people hear while the rabbis read; the people read while those with access to the printing press write; we write, while our techno-elite program. As a result, a majority of people remain one dimensional leap of awareness and capability behind those who manage to monopolize access to the real power of any media age.
— Rushkoff, Douglas. “NET EFFECT: It’s not too late for humanity to survive the digital” Arthur. 2009.
Today, most of us think of computers—and all of our digital devices—in terms of the applications they offer: “What does it already do” instead of “what can I make it do?” Likewise, instead of teaching computer programming in school, we teach kids how to use Microsoft Windows. This difference is profound. It exemplifies the core difference between a society capable of thinking its way beyond its current limitations, and one destined to repeat the same mistakes until it drives itself to extinction.
— Rushkoff, Douglas. “NET EFFECT: It’s not too late for humanity to survive the digital” Arthur. 2009.
1.
