Tulip mania and the stock market

It's the first day of the lockdown in London when I speak to Anna Ridler on Skype, and I’m granted a look inside her studio in the middle of the crippled British capital. She's always worked from home, so you might think Boris Johnson’s restrictions wouldn’t change her daily practice very much.

 But that's not the case. Spring is when Ridler presents her work at festivals the world over, and all those bookings are now falling through. I was originally supposed to meet her in Eindhoven at the STRP technology festival, but that didn't happen. She was also supposed to go to California for a presentation, but that has been cancelled, and she will also lose the money she would have received for the event – a personal disaster. Ironically, her work is about the evaporation of financial value.

She was going to show her installations Mosaic virus and Myriad tulips at STRP, and artificial intelligence plays a special role in both. If you know Dutch history, you’ll be familiar with the tulip mania of the Dutch Golden Age, when one bulb, especially of a striped variety, was worth the price of an Amsterdam canal house. It was madness, in other words. Some argue that this was the first economic bubble, and all bubbles eventually burst.

What many people did not know at that time, and what the hysteria highlights, was that the stripes on those coveted tulips were caused by an infectious disease, the mosaic virus. Ridler wanted to criticize speculative bubbles, and saw parallels between AI, tulip mania, and the cryptocurrency Bitcoin. She built an AI installation in which the price of one Bitcoin determines the growth of invented tulips in a video. The higher the price, the more stripes the tulips get, and an algorithm then turns them into flower still lifes.

Ridler chose AI because there is a strong link between AI and the concept of the work. ‘AI is also a speculative bubble,' she says. ‘For example, you can already buy AI toothbrushes, though they cost a lot of money. AI really is a buzzword at the moment, just like Bitcoin has been for some time now.’ In addition, AI has a technical connection with the way in which the flower pictures are created. ‘The bouquets you see on the still lifes never existed in reality,' explains Ridler. ‘They’re often flowers from different seasons. It's not that a painter actually sat in front of that particular vase of flowers to capture it; he or she used fragmented images to create a still life. Still lifes are an idealised version of what the painter has observed over time.  Exactly in the same way, AI works with the dataset I trained. AI doesn’t make copies of the information I put into the machine just to reproduce it. It extracts fragments and uses them to make new artificial flowers. All the fragments thus become a new whole.’

Ridler herself collected these fragments by hitting the road with her camera: she photographed ten thousand tulips while living in the Netherlands for two months. She says that the camera is a good comparison with AI, especially when it comes to the question of who is being creative here, technology or man. ‘When you take a picture, the camera does the work,' she explains. ‘There’s a lot going on inside the camera technically, of which you may or may not be aware, but these are actions that you don’t perform yourself. ‘You’re not the painter who puts the image on a canvas. But you’re the one who went somewhere with a camera, chose a scene to capture, and then framed that scene. After that, you’re also the one who chooses which of the pictures you want to keep, and which you don’t.’ She laughs. ‘Who’s creative then: the camera or the photographer?’

There’s also a lot of human labour involved in programming the AI, she emphasizes. Choosing what data sets to use and how to set them up requires human effort, and the AI then makes creative suggestions. But the AI is not aware of that. She mentions AlphaGo, the first self-learning computer to win a game of go by making a creative move. ‘A move that the machine made without being aware that it was a creative move. You need people to realize how creative that was. It's the same with the technique I work with.'

Ridler uses generative adversarial networks, or GANs: algorithms that create new derivatives, variations and perspectives based on existing data, just like people come up with new ideas based on existing ones. ‘The GANs I work with give me interesting suggestions. But I’m the one who judges whether something is interesting to me and whether or not I’m going to use it in my work. ‘It's not a question of I press the button and the computer does the work. I have an editorial role with the output of the machine: I select and organize. That's what I need as an artist, because the machine can't determine what's interesting,  whether something is strange and striking, or whether something resonates with an audience. Furthermore, AI is as good as the information with which you feed it. When you feed the machine with cat pictures, the AI machine produces a cat, not a vase of flowers.'

Ridler, who studied English literature at the University of York and previously worked on a project with Wikileaks, does not like to be pigeonholed as an AI artist. Although this group of people uses similar techniques, their work is too diverse to make comparisons, she thinks. Her interest lies mainly in data, and she often uses technology to unlock it and learn something about the world. ‘Every piece of data is a trace of human life,' she explains. ‘My challenge is to extract stories from that data that I don't encounter in other ways.’

Other artists use AI because they’re curious about the technique itself, but Ridler employs it only when there is a conceptual link with her work that helps construct the message or the thing she’s interested in . One of her next projects, in which she works with newspaper archives, is completely analogue. ‘Using AI takes a lot of energy,’ she says. ‘The energy consumption is comparable to having a kettle of water on the fire nonstop for two months. I think you should have a good reason for using this technique. I choose all the materials I work with very consciously.'

How does Ridler regard the fear that AI will replace people? Is that a real scenario in art? 'It depends on how you define art. If you think an Ikea painting is art, yes, then computer algorithms and AI can replace people perfectly well and produce beautiful pictures to hang on the wall at home. But that's art without meaning. Art that you encounter in museums cannot be replaced by AI. It’s art that needs people to give it meaning. Artists are needed for meaning in the creative process, and an audience is needed for the work to resonate or not resonate.’