I can almost hear the AI quantumly snickering amongst its collective self from here.
In a remarkable first, a fancy, old-looking portrait has been sold for more than half a million dollars. Clearly, that’s not the first. What’s new is that it’s the first art of its kind to land on the Christie’s auction room floor.
A fake picture, invented by AI algorithms, painted by a robot.
And to make it even more of a mess, it’s dividing AI art coders because the ‘artists’ behind this painting used algorithms from other AI artists to make it.
Nineteen-year-old AI artist Robbie Barrat tweeted up a storm as the sale progressed, showing how he had collaborated with the French art collective called Obvious, who are now half a million richer thanks to their access to these artistic algorithms.
Obvious didn’t exactly make it, well, obvious, that they were building their own work on the collective efforts of an open-source community of AI artists. But once it was raised as a factor they admitted they worked with Barrat up to a point, and, they claim, they have made their own tweaks to the code that produced their work. So it’s only based on his code, but the code is now different and therefore their own ‘work’.
But isn’t this fundamentally how art works? And isn’t Barrat essentially upset because someone from a happy-go-lucky community of friendly AI art coders has added some clever marketing to the idea and made some bank?
Does an art teacher get upset when the kids they teach go on to become commercially successful?
Barrat himself admits he’s upset because he thinks Obvious isn’t about art, it’s about marketing. But isn’t most financially successful art about how you sell it?
Obvious gave the painting a title, Portrait of Edmond Bellamy, and it’s from a series of AI portraits they have produced depicting a fictional family of Bellamy’s. It’s actually the most bland and boring image from the entire series.
So much of successful art is marketing.
The shredded Banksy is worth more now it’s shredded because it has a cooler story. Marcel Duchamp’s ‘Fountain’ was simply art because he said it was. He didn’t make it, he just had enough ego to put a urinal on a pedestal.
Andy Warhol was amazing as a marketer of his art. And indeed, Warhol had a print in this very same Christie’s auction – it went for one quarter of the price of the AI artwork. Take that, soup tin guy!
In this case, isn’t it just that the less-good-at-sales AI artists are annoyed because they’re being left behind by the very-good-at-sales AI artists who came up with better marketing?
In another sense, Barrat is a digital Frankenstein, and his algorithm has become his personal monster. He let it off the leash, and it left him behind. It grew up and found new friends. And those friends made all the money.
This style of AI is known as an ‘adversarial’ network. It exists in two halves. One, the Generator, tries to create something good enough to fool its other half, the Discriminator. This dynamic relationship makes the AI get smarter as it learns what does and does not fool its counterpart. It’s like an AI version of Battlebots.
But the AI has really won. What we’ve seen it generate in the end is a new adversarial network of AI artists. A battle of success and failure at convincing the wider art world that the work is real. Worthy of recognition, and worthy of a price tag.
Obvious got there first. And that first raises the bar for all those yet to jump over it. Creating a need to code ever smarter, iterate ever faster, to find a way to take their art to another level.
There’s only ever one first. And this first carries a signature at the bottom right of the artwork – a simple looking algorithm that represents an AI that just broke the hearts of those it left behind.