, Seoul Mediacity Biennale, Seoul Museum of Art, South Korea, to 18 November. Keep an eye on聽Culture for future showings
NINE portraits hang on a wall. The mouths and eyes have been scrubbed out, the faces painted over or dissected with thick smears of colour until the sitter sinks out of sight. This is resistance art: the revolution will not be recognised.
is an installation by South Korean artists Shin Seung Back and Kim Yong Hun. Commissioned for the Seoul Mediacity Biennale 2018, an exhibition at the Seoul Museum of Art, the work is one of many examples underlining just how complicated it is becoming to tease apart humans and the complex machines they create.
Advertisement
The duo set nine fellow artists a task: paint a portrait of Kim that a human would recognise as a face, but a computer running three different face-recognition algorithms would not. The project was an experiment, says Shin. Could the artists find the shrinking space that is still wholly human? 鈥淚f they failed, it would show the delicate situation we are facing,鈥� he says.
Face recognition runs through Shin and Kim鈥檚 work. A previous installation involved images of clouds in which software had found faces. A was shown at the UK鈥檚 AND festival in 2015, with cameras scanning the skies above Grizedale Forest in Cumbria to collect 鈥渇aces鈥�.
We are used to our mugshots being catalogued online. Facebook and Google use face recognition to tag people in uploaded photos. But video surveillance now comes with face recognition too. The crowds at Madison Square Garden in New York are scanned for known troublemakers. Cameras at pedestrian crossings in Shenzhen, China, watch jaywalkers.
Some people have taken to wearing obfuscating headgear, from hats and scarves to camouflaging face paint. But the latest algorithms aren鈥檛 fooled.
Of course, a portrait by Picasso or Francis Bacon would probably escape recognition too. But in Nonfacial Portrait, the process of avoiding detection is part of the work. Each portrait is exhibited next to a showing the its painter played with the algorithms.
A camera films the painter from above as they work. As soon as the computer spots something on the canvas it thinks is a face, a red, green or blue square (one for each algorithm) highlights the bit of the face it has identified. The artists react by painting over the recognisable features or adding blocks of colour elsewhere to throw the software off.
It wasn鈥檛 easy, says Shin. 鈥淚f you make the portrait close to the subject, machines will easily detect the face. And if you distort the face too much, the painting could not be seen as a portrait of the person.鈥� It is up to the audience to decide if the artists succeeded, he says.
鈥淎s soon as the computer spots something on the canvas it thinks is a face, a coloured square pops up鈥�
This year鈥檚 Seoul Mediacity Biennale is themed around the idea of living well, after the ancient Athenian notion of eudaemonia 鈥� the good life, or happiness. Since the exhibition explores how well this notion fits into a high-tech future, it is no surprise that many of the artworks deal with AI. Human well-being will largely depend on how we navigate our relationship with this technology.
Another highlight of the biennale is by Seattle-based artist Mike Tyka, who has a day job at Google. Tyka鈥檚 installation consists of 20 printers that spit out long strips of AI-generated political tweets by fake people with AI-generated faces.
He trained his software on 200,000 tweets from accounts that were found after the 2016 US elections to be run by bots. In the middle of the ticker-tape torrent of tweets are a pair of chairs, inviting people to sit and talk.
There is still fun to be had, though the line between humans and machines grows fuzzier. 鈥淚t will be more and more difficult to find unique human abilities as technology develops further,鈥� says Shin. 鈥淏ut we need to keep looking for it, not to find our supremacy over machines, but to know who we are.鈥�
This article appeared in print under the headline 鈥淗ow to paint over AI鈥�
