Thank you – we'll be in touch soon!

Back to Innovation

The Death of the (human) Author?

Screen Shot 2016-10-31 at 16.15.03

When I was a young, naive English student who thought she’d scaled the heights of literary study by having a well-thumbed copy of Shakespeare’s complete works and a big swooshy scarf, my mind was well and truly blown when I landed at university and read Roland Barthes’s The Death of The Author.

The title’s pretty self-explanatory, but here’s a summary (spoiler alert!): in his essay, Barthes basically says there’s no point trying to figure out what any writer intended in their work, because as soon as those words hit the page / vellum / MacBook, they’ve got nothing to do with the writer any more and are out there, on their own, open for interpretation. Ok, he doesn’t mention MacBooks — but other than that I think that’s what he’s saying. Apart from shattering my previous understanding of literary studies and seemingly rendering useless multiple visits to the Shakespeare Birthplace Museum, this essay changed the way I saw human interaction with art in general. Does what the artist intended actually matter? Is it possible to “miss the point” of art, or is the value entirely in our interpretation of it? And what do we even mean by “matter” and “value” when we talk about art?

Don’t panic, I’m definitely not going to even attempt to answer those questions in a blog post (I’m now an old, cynical, ex-English student and know better than that). But these questions all came back to me recently when thinking about the application of artificial intelligence in creating art — be that music, visual or literary forms. We’ve now seen AI making customisable soundtracks to video, deep learning composing jazz, computers impersonating impressionist painters, and an AI called Benjamin (sure) who wrote a short sci-fi film.

All signs point, then, to the imminent death of the author, artist or music-maker — and in a somewhat more literal way than Barthes intended when he wrote about it (although, what does that matter, eh?). But, is there something much more interesting going on here than the machines simply taking over? All the examples just listed have been “trained” by humans. They’ve been fed the equivalent of whole libraries of human creativity, and what they’ve provided us is a new pastiche of all that input. In 1967, Barthes pointed out this is how all new literature is formed — a collage, “made of multiple writings, draw from many cultures and entering into mutual relations of dialogue, parody, contestation..” — and, to bring my sources up to date a little, Patrick Stobbs, who founded AI composer Jukedeck.com, points out that “creativity is…new combinations”. This is something that humans are great at, but, as Patrick points out, it’s also something computers can be great at. Though, he is careful to caveat that it remains to be seen whether people “enjoy art as much” when they know it’s been created by a machine.

Which brings us nicely to the reader — or the viewer, the listener, the audience. In Barthes’s essay, the reader is king – she or he adds an entirely new, unique lens to the way any piece of art is consumed.Or, as put by Anastasia Leng — whose tech startup Picasso Labs uses image recognition and machine learning to predict which creatives will be more successful than others — “creativity, like all datasets, is in the eye of the beholder”. The reason Picasso Labs exists to provide insights and predictions to content creators is because, of course, the consumer is king — even more so now than when Barthes pointed it out. We all know products, ideas and experiences that thrive now are those that centre on making life brilliant for the end user. Similarly, whether developed through an AI or a human brain, a piece of any kind of art will only gain value when it’s engaged with by the end user.

Perhaps the killer app for AI in the creative world, then, is for it to help deliver more exciting creative to more recipients. Jukedeck has helped singer-songwriters who don’t play an instrument themselves to write songs with an AI-generated accompaniment so they can release their art to the world; Picasso Labs has helped brands to provide content that’s much more interesting to their specific audiences, cutting through the chaff.

The goal of huge leaps in technology is frequently cited as democratisation. Whether that’s an intended goal or a happy side-effect, maybe that’s where AI can help us get to in terms of opening up creativity — both in producing and consuming it. Barthes would approve — he reckons it’s the egotistical sense of the “prestige of the individual” (along with “capitalist ideology”, if you want to get radical about it) that led us to venerate the author for all those years. If the use of faceless, ego-less (at the minute, anyways…) machines can help make creativity more accessible, maybe AI won’t kill the author after all — it’ll just bring it to life in all of us.