Tweet
This essay covers a lot of ground discussing the various ways that art is transformed/mediated/produced/curated by algorithms, but I want to focus on one point in particular.
What we crave most in art, what we reward more than anything else, is surprise. Marcel Duchamp’s urinal, the introduction of perspective to landscape painting, stream-of-consciousness literature – these creative breakthroughs achieve much of their impact by shocking us into some new perspective on the world. Little wonder that the modernist poets were so fascinated by the metaphors of blasts and explosions, or that art has such a long and complicated history with warfare. We need art to surprise us in order to blow up the world, to create fissures out of which the new can emerge.
Computation is not good at this. Algorithms are wonderful for extrapolating from past information, but they still lag behind human creativity when it comes to radical, interesting leaps. So far, they are much better at identifying and replicating surprising content than they are at producing it themselves. Platforms such as Facebook or Flickr’s ‘interestingness’ quotient ultimately measures a kind of surprise, one that draws on information theory as well as aesthetics. We respond to viral memes on social media because they produce something unexpected, often leveraging the deep relationship between surprise and humour. It is telling that so many memes now hide their linguistic tells (more tractable to algorithmic watchdogs than images) inside GIFs and JPEGs that circulate in a kind of shadow economy of surprise.
Surprise will remain a human territory, at least for the short term, because it is so idiosyncratic in the first place. Our sense of the unpredictable is so oddly tuned that true randomness can sometimes seem too regular, too predictable, like a long string of coin tosses where the same side comes up many times. At the same time, we are quite choosy about the kinds of novelty that count, a form of distinction that could, in the end, be precisely what we mean by aesthetics. How many art critiques and book reviews boil down to the judgment ‘this is a predictable extrapolation’? Newness is necessary but not sufficient for human surprise. There is a cadence, a significance that we seek in the aesthetics of surprise that reaches deeper than mere randomness. As pattern-seeking animals, we are looking not just for comprehensible behaviours but for signs and portents – stories about the world that allow us to configure reality according to an aesthetic logic.
Computation is not good at this. Algorithms are wonderful for extrapolating from past information, but they still lag behind human creativity when it comes to radical, interesting leaps. So far, they are much better at identifying and replicating surprising content than they are at producing it themselves. Platforms such as Facebook or Flickr’s ‘interestingness’ quotient ultimately measures a kind of surprise, one that draws on information theory as well as aesthetics. We respond to viral memes on social media because they produce something unexpected, often leveraging the deep relationship between surprise and humour. It is telling that so many memes now hide their linguistic tells (more tractable to algorithmic watchdogs than images) inside GIFs and JPEGs that circulate in a kind of shadow economy of surprise.
Surprise will remain a human territory, at least for the short term, because it is so idiosyncratic in the first place. Our sense of the unpredictable is so oddly tuned that true randomness can sometimes seem too regular, too predictable, like a long string of coin tosses where the same side comes up many times. At the same time, we are quite choosy about the kinds of novelty that count, a form of distinction that could, in the end, be precisely what we mean by aesthetics. How many art critiques and book reviews boil down to the judgment ‘this is a predictable extrapolation’? Newness is necessary but not sufficient for human surprise. There is a cadence, a significance that we seek in the aesthetics of surprise that reaches deeper than mere randomness. As pattern-seeking animals, we are looking not just for comprehensible behaviours but for signs and portents – stories about the world that allow us to configure reality according to an aesthetic logic.
The author argues that good, revolutionary art is predicated on the ability to surprise us and reshape our worlds, and that all the finely tuned algorithms that exist (so far) are not very good at this. So my question is, for how much longer will that be the case? Is art so deeply human that it will never be ceded entirely to computers? Or will there come a point at which competing against artistic AI is fruitless, much as we would never ask a human to physically build a car?
Comment