48 FUTURE SHOCK Artificial Intelligence
THE POWER OF MACHINE LEARNING AND DATA SCIENCE TOOLS AND TECHNOLOGIES IS HELD BY A SMALL SELECTION OF COMPANIES
A PURITANICAL AI
FINDING FAIRNESS Stuart Coleman prefers not to name “an organisation that developed a computer vision algorithm that could look at things and infer lots of knowledge”. “Their mission was to share machine learning algorithms for the benefit of society – and they realised that the most compelling use for this was really underworld stuff. In a society where it’s deemed OK to be watched, people are living in a Minority Report situation. You cannot go into most urban areas in China without the state knowing where you are, who you’re meeting... It’s very easy to be unaware of the influence these platforms have. There are very few sanctuaries to escape this stuff, because we all crave the convenience.” Machine learning has been applied to fields as diverse, and as impactful, as commercial health insurance, business
Lotte Louise de Jong is exposing impressionable minds to pornographic images.
De Jong and a programmer used open source written code fromNeuralTalk and an open
The Dutch media artist, in her short film Talk Neural To Me , ‘showed’ a porn video to an AI and asked it to describe what it saw. The AI, like a shocked human entirely out of its depth, could only offer up scenarios based on its limited knowledge of the world. De Jong’s film consists of blurred footage sourced from PornHub, accompanied by a text-to-speech voiceover of the AI’s assessment of what it thought it was seeing. The results are by turns hilarious, pathetic and frightening. Exposed to sex acts, the AI offered descriptions like: “a woman in a red shirt is holding a remote”, “a woman is brushing her teeth with a toothbrush”, “a person is laying [sic] on a bed with a stuffed animal” and “a woman is taking a selfie in a mirror”. Some of the descriptions are abstract enough to be Freudian poetry: “a close-up of a stop sign on a wall” and “a view of a city with a clock tower”. And of course, there is the inevitable “a person is holding a banana in their hand”. “It was the first time I worked with neural networks for a project,” says the artist. “It’s really interesting to use it as an art form, because you can create this distance between viewer and subject.” De Jong made the film in 2016 during her degree, when easily accessible neural net technology was still relatively new. “What if the computer were a person? Howwould they look at this? It doesn’t have an opinion or personal experience.”
source database. The neural network analysed images from the films grabbed every 15 frames or so. “There was also an element of humour,” she says. “It couldn’t really read what was happening. What struck me is how limited it was, actually. It is smart in some ways – it could describe the action very well. Of course, it wasn’t someone ‘brushing their teeth’, but just based on a still, it could still understand the movement.” She was particularly struck by how the AI wasn’t able to comprehend naked images. “Even in a healthy, normal way, it just doesn’t read that. Anything that had something to do with sex or nakedness was transformed into this really abstract thing. After a while it got repetitive.” Most AIs, as they are currently trained, aren’t very good at describing humans without their clothes on. In those data sets, a human is a thing that has clothes on it. These smooth-surfaced objects with no recognisable sleeves, trousers, shoes are, to the AI mind, mysterious and alien. Machine learning is employed more widely now for purposes of content management and censorship, but these tools are still quite simple, and do no more than detect whether clothes are present. “We also, as a society, think of sex and pornography as something very much apart fromwhat we see normally,” notes de Jong. “It’s taboo. And so within the neural networks, they then also become a taboo.” See de Jong’s work at lottelouise.nl
information and politics, but perhaps the most concerning developments have been in criminal justice. As the University of Southampton’s Chapman
explains, parts of the US court system began using
feedzinesocial feedmagazine.tv
Powered by FlippingBook