Robot-Generated Art Is Going to Destroy Culture

Writers seem to have an ingrained fear of the idea that books might be written by machines. In Orwell's 1984, books are produced in a novel-writing machine "like jam or bootlaces." Meanwhile, the glut of recent films in which human men are shown wanting to fuck what is in effect a sexy lady toaster ( Her, Ex Machina, Chappie, etc, etc) all at one point or another raise a question that's taken to be some kind of mind-blowing blasphemy: could a machine ever create a poem, or a film, or a piece of beautiful music? While it speaks to a certain anxiety surrounding the increasing automation of all areas of life, the question is actually profoundly stupid. Machines are already creating our art for us. The real question is whether it's too late for any of that art to be any good.

Take, for instance, Amazon's recent announcement that it will start paying royalties to authors based on the precise number of pages that are actually being read. As several frothing commenters on news reports have pointed out, this move isn't entirely what it's been made out to be. The switch only applies to self-published authors whose ebooks are available on the site's monthly subscription service, and it's meant to correct a previous system which was felt to unfairly penalize longer texts (before, royalties were simply paid whenever 10 percent of any book had been read). Like all pedantic objections, this one doesn't really matter. The precedent's been set, and the global rollout of pay-per-page is now a looming, monstrous inevitability.

This should worry you, even if most of your income doesn't happen to come from your series of self-published erotic Bigfoot tales. Some people might find it disquieting that Amazon is tracking exactly what they're reading and exactly how much of it they finish, at once fulfilling the roles of a paternalistic school-teacher and a terrifying, blank-faced secret policeman—but, to be honest, what did you expect? Everything we do now is added to the vast reams of data that's slowly suffocating our planet with its ever-increasing weight.

None of this information belongs to the person that it's ostensibly about. When you die, bailiffs will probably show up to your funeral to collect your body for some awful tech startup, brutally referring the mourners to a clause hidden in a licence agreement you never bothered to read. There's nothing you can do about this, and you may as well accept it. The privacy concern in the end reveals a deep narcissism: the horrible truth is that as an individual you don't really matter at all, and nobody really cares what books you've been reading.

What we should be worrying about is what all this will do to literature. For centuries, great texts have been almost entirely subsidized by people who buy them out of a vague sense of intellectual guilt, read exactly one half of the first chapter, and then never touch them again. Nobody in all of human history, for instance, has ever actually read Finnegans Wake. Only madmen and ministers have read all the near-identical Synoptic Gospels from end to end. Really important books are not the ones that whisk you off into some escapist fantasy where your greased attention slips from paragraph to paragraph as you grin like an idiot child on a playground slide. They bear down on you with the solemn, ugly weight of duty.

Under the pay-per-page model, a vast spider's-web algorithm will eventually collect all the details of every ebook reader on the planet. It will know, for every book ever written, exactly where most readers gave up, how long they took to try again, and where they finally abandoned it forever. It'll know which chapters were read and reread, and which ones a hyperactive, screen-dazed public shamefully skipped over. And, when the difference between fame and starvation rests on how long you can grab an increasingly distracted population, eventually someone will come up with a mathematical formula to create the most profitable possible book. When that happens, there'll be no more writers. Computers and algorithms will be the ones actually writing books; human beings will just arrange the words. It's possible that these books might even be good. But it's very, very unlikely.

In fact, something much like this has already happened. When it created its remake of House of Cards, Netflix scripted the show according to its vast bank of user data. The company collects over 30 million data points a day—it records exactly when in a show you choose to take a toilet break; it can work out, based on timestamps, if you've ended your box-set binge out of actual boredom or because the sun's coming up and the gentle pong of stale pizza has become almost tangible.

As far as Nextflix is concerned, you're not just a passive consumer, but you are essentially a moron. This is why the emergence of "Netflix and chill" as the 21st century's go-to euphemism for sex doesn't bode well for the next generation's chances. It was on the basis of all this data that House of Cards was created, intended to be the best, most compulsively watchable TV show ever.

You can like House of Cards if you want; for the time being at least, you're still (just about) entitled to your own opinions. But it's not very good. The show is a cretinous and ungodly chimera, full of spiralling, pointless subplots, desperate to impress on the viewer what an unpleasant person its main character is, and bafflingly over-reliant on weird nautical metaphors ("It's how you devour a whale, one bite at a time," or "I love that woman more than sharks love blood"). In recent seasons it's whittled away any of the moral complexity it once feigned, turning itself into something like the West Wing with an even duller opening sequence—presumably at the behest of the machines.

The algorithmic novel can only be something similar: a big disjointed mess of genre clichés and patronising bitesize bit-words, where every chapter ends in a cliffhanger and everything is utterly, hideously homogeneous. A book of averages for a reader without qualities.

It's not that books written by machines are necessarily bad. In the early 20th century, the dada and surrealist movements experimented with automatic writing. They thought that by writing without thinking, as a purely mechanical exercise, the unconscious mind would reveal itself on the page. Gertrude Stein took things further. Her experiments in motor autonomism weren't designed to reveal anything, but to produce a pure writing free of meaning. Revolutionary composers following Schoenberg have used mathematical set theory to compose orchestral pieces. They thought they were creating something liberatory and avant-garde. It's not the machines' fault that we prefer worthless pap.

The problem lies with us, the humans. The robot poet of the movies first needs to be programmed with an understanding of what a poem is before it can write one, and that understanding is usually based on what filmgoers read. That's why when these fictional machines do create something, it's usually a boring representative watercolor, or music full of the stupid sweeping violins that people think of as being "beautiful." Technology just creates an average; the nature of that average is down to us. The only way anything could be salvaged is if enough people get rid of their e-readers and go back to blind, dumb paper. In other words, we're doomed.

Follow Sam on Twitter.

More from VICE:

Britain's "Sharing Economy" Is Creating a Desperate Servant Underclass

A Year On From His Kidnap, a British Man Is Still On Death Row in Ethiopia

FEMEN: Sextremism in Canada