AI is a concept turned reality. It is inescapable and is having a pervasive effect on our everyday lives. For this reason, we need to keep interrogating AI’s definition, purpose, the drive, the benefit and where it might fail us as a society fundamentally built on connection and community. We are nothing without each other. It’s not our first rodeo writing and thinking about AI, it can’t be. It is an ever evolving medium that demands constant reflection and analysis, to harness it’s power for good and remain sceptical to it’s force.
The Art Show on Radio National gave Faster Networks pause for thought, as did Ezra Klein’s writing and podcasting recently. Ezra has many questions, most of them unanswered despite making concerted efforts to immerse himself in AI culture, literally moving to the Bay area of San Francisco. He puts questions directly to the makers, the developers and investors and he wonders, ‘does AI have the power to manipulate us? and if not now, when?’ The Art Show asks is AI capable of using it’s own imagination? If not now, when?
A Chicago construction worker this week used the art generator, Midjourney, to create what he thought was a harmless image of the Pope wearing a ridiculous and outlandish but also right on trend, white puffer. I forwarded it to friends with a lavish compliment for the Pope, celebrating his iconic style. It was a fake. No surprises this image went viral and for a writer well classed in deep fakes and imposter art, I wondered, ‘where was my digital scepticism?’ Scepticism is more important than ever at this time, especially where real human figures are being used, as a joke or otherwise, to manipulate reality. The creator of the Pope pic was banned from Reddit and believes using alive, public figures is probably where the line should be drawn.
Artists especially, are being used by image generating platforms, such as Midjourney, to create styles of art in the vein of actual living and working artists. As Kim Leutwyler explained on The Art Show, these platforms require data points to provide users of image generators access to certain styles. She could write a prompt, such as, “in the style of Kim Leutwyler” and something significantly similar to her work would show up. I tried the same thing with a few different artists and had the same result. It looked like the art had been lifted from a gallery website or the artist’s page and then included other objects or patterns as described. Leutwyler wondered how Midjourney or DALL-E 2 were accessing these images and she found a website that you could check if your work had been used to train AI software, Have I Been Trained? This is a website where the user types in any name and a search of 5.8 billion images will let you know if your work or your face has been used to train AI software to create new images. Many a Pope over the ages are all heavily featured in a search on Have I Been Trained?.
Leutwyler didn’t seem bothered by the democratisation of her creative work, outside of the obvious violation of not being asked. Leutwyler is a portrait artist who wants to paint as many humans as possible and time is limited and, the cost of her work is inaccessible to many, she is an Archibald finalist afterall. However, she also doesn’t want her name attributed to art that she didn’t create.
Analysing AI through pop culture and creativity seems a good entry point but surely it is only surface investigation. Digging deeper into AI reveals unregulated chaos where the power is in the hands of a few. Humans are struggling to imagine what the future looks like with AI at the helm but as the race to develop tools that are ever smarter than us continues, our imaginations need to expand and quickly.
