“assorted still lives of nothing in particular” pt 2
in order, “a fruit and some bones”, “all quiet on the western front”, “a bouquet of colorful flowers”, “a bowl of fruit that looks like a mountain range”, “a concrete megastructure”, “an acropolis landscape”
“Do you know what an emissarial projection is…? It’s when they send a tuned suite of queries and responses to a remote location, by light beam. To play the part of an emissary.”
“…If we are to believe what we are told, the object they send is something like a library, transmitted by a signal laser. Suitably housed and emplaced within enabled equipment of sufficient capacity and complexity, this… entity, though it is simply a many-branched array of statements, questions and answers, with a set of rules governing the order in which they are expressed, is able to carry out what seems like an intelligent conversation. It is as close as one is allowed to to come to an artificial intelligence.
I recently started playing with DALL-E 2, which will attempt to generate an image to go with whatever text prompt you give it. Like its predecessor DALL-E, it uses CLIP, which OpenAI trained on a huge collection of internet images and nearby text. I’ve experimented with a few methods based on CLIP, but DALL-E generates particularly clear, coherent images.
So of course I decided to use it to mess up corporate logos.
ALT
“The local Waffle House” - generated by DALL-E2
ALT
The Pizza Hut logo - generated by DALL-E2
ALT
The Applebees logo - generated by DALL-E2
ALT
A sign for In-N-Out, viewed from a distance - generated by DALL-E2
AI-art confronts us with a truth we might prefer to deny: human-made commercial art has long been “inhuman,” because it was tailored by and for the ever-more-specified demands of the market. The artist was just a set of hands operated from on high by what was almost already an algorithm of the if-you-liked-this-then-you’ll-like-more-of-the-same variety. I think of one of the pulp writers who would bang out a novel a week by consulting the plot chart tacked above his typewriter, itself presumably based on what had already worked; for an updated reference, think Save the Cat! And a lot of the pleasure serious audiences—fellow artists, critics—have always taken in mass art comes from detecting signs of the artist’s irrepressible spirit in the otherwise automated production, i.e., the human touch, what the famous auteur theory was developed to describe in the case of commercial cinema.
But then look at modern high art, its more and more desperate, strenuous, and indeed absurdist evasion of the “word coined by commerce”: eliminate depth, eliminate sense, eliminate human interest, eliminate humans, or so says the avant-garde, and then implement one or another formal protocol—Impressionist, Cubist, Fauvist, Imagist, Suprematist, Abstract Expressionist, Serialist, et al.—to make art in the absence of either organic mimesis or organic self-expression, lest you be suspected of a commercial appeal. So the work the avant-garde produced was inhuman too, less human than some of the mass culture they fled so fearfully.
Not to mention academia: whether formalist or historicist, whether regarding the text as an impersonal freestanding structure whose origin is of no concern or as an impersonal social site where ideologemes converge, the scholars professionalized their disciplines by refusing to consider the objects of their study—works of art—as anything so unscientific as the products of individual consciousness.
Two of Hoel’s sources, Benjamin and Tolstoy, are unreliable witnesses for the humanistic defense of art; their own theories lead to art’s automation. The Marxist Benjamin was not lamenting the loss of aura; he was hopeful about the democratization and politicization of art it portended. Similarly, Tolstoy is a forerunner of socialist realism when he claims, in lines Hoel quotes, that the artist “should stand on the level of the highest life-conception of his time,” i.e., should transmit the wisdom of the collective, not the individual consciousness, wisdom that might as well be automated and programmed. Only John Berger among Hoel’s authorities makes the strict case that art, to be art, must be the product of the individual, though here his modernist sentimentality is somewhat at odds with his Marxism (and so much the worse for his Marxism).
And I’m not assigning blame for all of the above, for the modern inhumanism: art really is the place where the human touches the inhuman, where individual consciousness must mix itself with recalcitrant matter and with the calcified social to produce new configurations and totalities. To value this transaction most for what it tells us about individual consciousness is a choice, one I agree with Hoel that we ought to be making, and ought to have made sooner, but one that can’t be reclassified as other than a choice by playing with the definition of art. I would go further and say that in the age of AI we will simply have to know whether a given work of art is or is not human-made, how and to what extent, and to decide to value it more if it is.
We should return to the possibility of being moved by inhuman art when we know it was made by human minds and human hands, even if the artists toiled in a commercial cage or reacted so violently against this imprisonment that they caged themselves some other way. This cage or that, we’re capable of being moved all the same before a Jackson Pollock or a Jack Kirby, before a Samuel Beckett or a Lana del Rey. But that’s because we know someone’s in there, in the one cage or the other, a live soul beating wings against the bars.
If we don’t know, will we respond the same way? And can we tell just from the surface of the work? Just by looking? If you’d never read Tender Buttons before and I showed it to you and said an AI wrote it, wouldn’t you believe me? And yet when you know an AI didn’t write it, when you find out what a fascinating character composed those lines, aren’t you—not me, I never finished that book, but you—capable of being moved? So knowledge matters first: a human being made this. After that, belief: a human being isn’t just any kind of being. The soul is never a question of evidence but always a leap of faith.
I recently started playing with DALL-E 2, which will attempt to generate an image to go with whatever text prompt you give it. Like its predecessor DALL-E, it uses CLIP, which OpenAI trained on a huge collection of internet images and nearby text. I’ve experimented with a few methods based on CLIP, but DALL-E generates particularly clear, coherent images.
So of course I decided to use it to mess up corporate logos.
ALT
“The local Waffle House” - generated by DALL-E2
ALT
The Pizza Hut logo - generated by DALL-E2
ALT
The Applebees logo - generated by DALL-E2
ALT
A sign for In-N-Out, viewed from a distance - generated by DALL-E2
We love how literally the AI took our IF title It looks so cool, like deep blue veins; thank you so much for including ours in this beautiful compilation! Check out OP’s other AI-generated IF renderings, they’re so interesting!
ALT
ALT
ALT
ALT
ALT
ALT
