Last time, I wrote about the Library of Babel– once a short story by Jorge Luis Borges, now an online library of every possible combination of text, divided up into pages of 3,200 characters each. Therefore, the library already holds every possible piece of writing before it is ‘created’ by us. That includes every poem, script, story, even this very article itself.
However, as I mentioned before, this prospect has been generally met with apathy. As the library contains everything, the numbers of useful, interesting or valuable pages are lost in a tremendous volume of inane nonsense. Furthermore, the library is almost entirely unable to recognise the respective value of its own contents. It has a basic dictionary function and so can distinguish between words and random strings of letters, but that’s about it. Therefore, to access any specific writing within the library you have to generate it yourself, arguable negating the library’s value as an artistic resource. The same is true regarding its image catalogue, so it would seem that the library does not pose a threat to your creative talents!
And what a relief that is. As humans, we’re resigned to the fact that computers will always surpass us in the realms of cold, unemotional mathematical calculation. Yet we like to think that creativity and the production of artwork is something unique to humankind and thus beyond the reaches of artificial intelligence.
However, perhaps we shouldn’t be resting on our laurels. Just because the Library of Babel is unable to recognise the significance of its own contents, this doesn’t mean that there aren’t artificial programmes which can. Take Google’s Imagenet, for example– a software which has learnt to recognise the contents of images with frightening levels of accuracy. Incorporate an Imagenet-powered search function with the Library of Babel’s vast collection and maybe we’ll be pulling perfect logos from its database before the graphic designer even has the chance to turn on their Mac?
Returning to words, and the original focus of the library, Google has also been feeding it artificial intelligence romance novels, with the hope that it will eventually garner a sufficient enough understanding of the language to produce its own literature. Though a little more distant, how far away are we really from a world in which a computer can generate a novel to rival that of Mills and Boon?
Still not convinced? Then check out this podcast from 2007 which features EMI, a computer programme producing its own Bach-esque symphony. The result sounded so authentic it convinced a lifelong Bach aficionado that it was an original piece.
But what does this all mean for the future of creative consumption? Computer programs may begin to develop their own images, music and novels, but will these ever truly be accepted by the general public? Surely we will remain loyal to the creative efforts of our fellow fleshy humans over any mass-produced, artificial art? Perhaps not, and if you’ll permit me a slight transgression, I’ll explain why.
The majority of people who spend their weekends browsing the internet for cheap deals on indie computer games will have come across the term ‘procedurally-generated’. Procedurally-generated games are those which contain an algorithm to produce a set of unique levels each time they are played. The level design must conform to certain rules, allowing the game to be playable and consistent to a degree; however, every player will experience something new and specific to them whenever they boot the game up. Take, for example, Spelunky. Billing itself as a ‘randomly generated action adventure’ Spelunky tasks the player to navigate a series of procedurally generated caves, filled with traps and monsters. Each time the player enters a new cave, the layout may be more-or-less favourable to them, creating a pleasing element of chance and novelty which adds to the game’s addictive qualities.
One of the advantages of procedural-generation is that it allows the production of large amounts of unique content very quickly. Can the same process be applied to artistic endeavours? The random level generator in Spelunky contained parameters which were pre-defined by the game’s creator, Derek Yu. Yet when considering the computer generation of art, surely giving the consumer more power to set their own parameters for the creative output is the lucrative route to peruse. Returning to the example of EMI, when EMI was fed Bach she created Bach-like music. Feeding her Mahler produced something akin to Mahler. Can we imagine a future where I feed software a series of Flannery O’Connor short stories and request “More like this please”?
I don’t think it’s too unlikely. Especially within the context of this recent Guardian article which suggests we’re increasingly looking to art to fit our own preferences and keep us within our comfort zone. Perhaps in the future, technology will be so advanced we need not even provide source material to draw from, simply asking the computer to create something which matches a list of our specific whims in that moment. At least, I certainly hope so. Until that day, if I want to read any more Benedict Cumberbatch vampire steampunk slash fiction set on Mars, I’m going to have to keep writing it myself.