I’m revisiting my thinking about what constitutes Art in the age of the Art Engines, and an open question for me as to whether there is a uniquely human component of this. Specifically i am responding to challenges and ideas from a few key collaborators and thinkers. Today i will switch my perspective: let’s accept AI as an authentic tool for creativity (artists using AI), and also as an independent agent of ‘creativity’ (AI churning out ‘output’). So what makes art?
A lens to explore today is intentionality: essentially, does intent matter? Do you have to intend to make art for it to count, or may you (or it) be unintentionally artistic?
In the previous piece, i asked about a graffit robot: trundling around, using it’s digital senses to find an alleyway, then spraying a message you have texted it, or maybe just independently choosing a stock phrase, printing a poster, and pasting it up. Is that art?
If ‘intentionality’ is important, then the first example counts as art (i intended to direct the robot to be creative) but the latter is not (the robot is just randomly pasting stuff up) – although one could argue that the person who programmed it ‘gave’ it the intention.
A different example would be stones on the foreshore: as i walk down the beach, i see a spread of stones on the sand, which resembles a galaxy cluster. But i don’t consider it art: it’s a random act of nature – a product of a chaotic tidal system – and my brain, conditioned to find patterns, has found one. So the sea is not creative, it has not made art. But if i take a photo of it, then i have. Through my attention, and intention.
Similarly, if i imagine those stones, but never actually see them, and i paint them, then i have created art.
What if Donald feeds both my photo and painting to a generative system, which then creates ‘art’. Well by the intentionality argument, then he has created an authentic piece of art. Note ‘he’ not ‘it’.
Ok, so in these examples it is relatively clear that we can trace the output back by one or two steps to the intent of the human: the photographer, the programmer, the wanderer on the beach, paintbrush in hand.
How far back can you abstract the intention for it to no longer count?
Chaos theory would indicate that when my son went paddling in the sea he changed the system, so has he got an authentic role as artist in the scatter of stones? I assume not or else everything would be art. If art is a subset of the whole, there must be a boundary somewhere. Also: he did not intend it.
But what if i go paddling, with the specific intent of changing the system to create art? Are those stones now my authentic output?
Mark questioned me on this: what about AI artists, who use the AI generated content as part of a creative process? Well this would clearly count as authentic ‘art’ in this sense, as they intend it to be art, and are using AI as a tool.
More nuanced may be the question, what if the AI recruited the artist? It’s not inconceivable: i create the system, teach it about phone books, let it loose on Google, and it tracks down Mark’s AI artist friend, emails them an invitation (written by a narrative engine) and even pays them to paint. Who holds the intention? The programmer, the AI, or the artist who is being paid? Or is it not art now?
This model would reduce art to utility, like brick laying.
I guess we could carry this one further: imagine that we deny that this output is art, but the narrative engine then ‘writes’ a story of how upset it was not to be considered an artist. Is that literature? And if we fed that into an Art Engine, is that art?
Let’s take a different stance: religious iconography. Depictions of the prophets are forbidden in some interpretations of Islam. Would that count if the image was AI generated, and if so, who has been blasphemous?
This argument veers towards the view of ‘art is what i consider art’: so although there was no intention to create art, possibly the output is considered art, and hence offensive?
In one of the Identity Project interviews (currently unpublished) a Somali woman talked to me about the patterns of her headscarf and the meaning that they carried for her. Some interpretations of cultural appropriation would indicate that i should therefore not make wallpaper out of those patterns or images, because they have an origin, a cultural ownership, and should be protected. So can AI systems appropriate culture? Or not, because they have no intention?
What if i programmed a malicious Art Engine specifically to generate ‘art’ from indigenous Canadian artistic input. And to put that for sale on eBay as art. What if i programmed it to flood the market, hundreds of thousands of images. Would my culturally violent output be art? Or because it came from the AI, is it simply art, or noise?
Intentionality may be a valid frame through which to explore the evolution of art, but not without it’s own issues.
Again: i feel no obligation (as an artist or as a thinker) to land in a specific place on this with any speed (although as Mark says, this is the time to be considering it).