by Vanessa Kisuule, Temple Quarter Enterprise Campus artist in residence and Bristol City Poet
What follows is a strange and erratic collection of thoughts to accompany a poem I have written in response to a bot generated poem ‘written’ by Tiny Giant. The bot had the entirety of my last poetry collection A Recipe for Sorcery, a few of my favourite poems from other writers and some tech related writing provided by the University. The result was somewhat like looking at myself with all my body parts in the wrong place, toes on my forehead, elbows and knees swapped round. It was confusing, slightly sinister and endlessly fascinating.
The echoes of my work were immediately clear. They were not distorted enough to be unrecognisable, but there was enough deviation to intrigue and unsettle me. Some of the imagery jarred. Some of it occasionally touched on the profound. It was impossible not to wince at my own stylistic tropes: imagine having a friend write a parody of your work and reading it back to you, but with none of the tact that a human might offer! I’d recommend this process to any poet, but it would be remiss for me not to acknowledge the existential crisis it may cause.
Poetry is an artistic hinterland where abstraction and ambiguity can live more comfortably than in other forms. Linear narrative and logical coherence do not matter in this context as they would for a novel or a screen play. Poetry, then, may be the medium that AI could find the most ‘success’. Moments of incidental brilliance in the poem were not frequent, but there were enough to make me sit up. Some images genuinely moved and excited me; there was even the occasional jolt of envy at a particularly evocative turn of phrase.
It was hard to know what to do with these mixed feelings of admiration and disquietude. It’s rather foolish, and pointless, to be envious of an algorithm. It has simply done what was asked of it, deducing an end goal through its encrypted set of logical rules. Surely art is as much about intention and communication as the end product? Unlike you or I, it is unmotivated by a desire to connect with others or even a compulsion to sound clever, original or superior. And yet….
It’s a small comfort that it cannot produce anything without the initial input of words, but could the same not be said for us? Aren’t all the years of education, reading, talking and listening we do not our own form of data collection? The methodical approach of the bot may seem antithetical to the elusive creative process we purport to undertake, but this project has forced me to reconsider how we talk about the mechanics of art.
A huge amount of art has been produced through formulas for centuries, culminating in what we may loosely call ‘popular culture’. There are the four chord progressions and seven story formats that recur again and again in music, literature, film and theatre. More often than not, we pull from a series of long-established methods which we may copy or use as a spring board to ‘new’ ideas. Though we’re loathe to admit it, we, too, are algorithmic machines. AI may be the impetus we need to push ourselves out of our own tropes, or at least be more aware of how much we rely on them.
Marcus Du Sautoy’s timely new book The Creativity Code details how the new innovations in algorithmic learning are expanding our notions of whether a computer can ‘think creatively’. He talks about ‘bottom up’ code: algorithms that learn through data input and continually adjusting its methods for better results. In this way, the AI bot Deep Blue beat world champion Garry Kasparov having played millions of iterations of the game, learned from its mistakes and developed creative strategies to ensure a higher chance of winning. The bot can even make choices that have not been explicitly programmed by the writers of the code. It’s hard to overstate what a game changer this is: the potential for AI to innovate rather than regurgitate is ever more plausible.
The neural network used to create my poem also exhumes ghosts. In PLaiTH, Tiny Giant worked with other agencies to not only produce work in Sylvia Plath’s style but in her handwriting. This detail pushes at new level of invasive imitation, an unwitting mockery of penmanship and its implied intimacy. The project demonstrates how easily this can be learnt, aped and mass produced. Plath’s death is mired in controversy and the seductive myth of a genius’ life cut short. It’s natural to wonder what she may have produced had she lived longer. The Plath Project could be framed as a respectful homage, a creative continuation of legacy forming an ellipsis where once there was a full stop. Still, there is a difficult conversation to be had around assumption and ownership, on how we justify the use of intellectual copyright after death.
There is an online quiz called Bot or Not with fifteen poems in it which asks you to guess which were written by humans and which by bots. I got eight out of fifteen. It’s safe to say that the ‘humanity’ we assume is inherent in human writing is not at all obvious. Many people feel cheated when they learn that a piece of art that moved them was not made by a fellow human. Perhaps we shouldn’t be so quick to admonish ourselves, or demonise the AI. Our minds are built to create meaning and relevance in everything we consume. It’s why we anthropomorphise animals, why we tell stories and build religion. In a world where bot generated art is an ever increasing part of our reality, would it not be a natural extension of our survival instinct to engage with it? Ultimately, if the end result stirs you, does it matter who or what made it? I still don’t know if it does, but I’m sure our morality and imagination will expand to fit around this brave new world.