Alex Reben’s work is commonly absurd, typically surreal: a mash-up of large ears imagined by DALL-E and sculpted by hand out of marble; vital burns generated by ChatGPT that thumb the nostril at AI artwork. However its message is related to everybody. Reben is within the roles people play in a world full of machines, and the way these roles are altering.
“I type of use humor and absurdity to cope with a number of these points,” says Reben. “Some artists might come at issues head-on in a really severe method, however I discover when you’re a little bit absurd it makes the concepts extra approachable, even when the story you’re making an attempt to inform could be very severe.”
Reben is OpenAI’s first artist in residence. Formally, the appointment began in January and lasts three months. However Reben’s relationship with the San Francisco–primarily based AI agency appears informal: “It’s a little bit fuzzy, as a result of I’m the primary, and we’re figuring stuff out. I’m in all probability going to maintain working with them.”
In truth, Reben has been working with OpenAI for years already. 5 years in the past, he was invited to check out an early model of GPT-Three earlier than it was launched to the general public. “I received to mess around with that fairly a bit and made a couple of artworks,” he says. “They have been fairly all for seeing how I might use their programs in several methods. And I used to be like, cool, I’d like to attempt one thing new, clearly. Again then I used to be largely making stuff with my very own fashions or utilizing web sites like Ganbreeder [a precursor of today’s generative image-making models].”
In 2008, Reben studied math and robotics at MIT’s Media Lab. There he helped create a cardboard robotic known as Boxie, which impressed the lovable robotic Baymax within the film Large Hero 6. He’s now director of know-how and analysis at Stochastic Labs, a nonprofit incubator for artists and engineers in Berkeley, California. I spoke to Reben by way of Zoom about his work, the unresolved stress between artwork and know-how, and the way forward for human creativity.
Our dialog has been edited for size and readability.
You’re all for ways in which people and machines work together. As an AI artist, how would you describe what you do with know-how? Is it a software, a collaborator?
Firstly, I don’t name myself an AI artist. AI is solely one other technological software. If one thing comes alongside after AI that pursuits me, I wouldn’t, like, say, “Oh, I’m solely an AI artist.”
Okay. However what’s it about these AI instruments? Why have you ever spent your profession taking part in round with this sort of know-how?
My analysis on the Media Lab was all about social robotics, taking a look at how folks and robots come collectively in several methods. One robotic [Boxie] was additionally a filmmaker. It mainly interviewed folks, and we discovered that the robotic was making folks confide in it and inform it very deep tales. This was pre-Siri, or something like that. Nowadays individuals are conversant in the concept of speaking to machines. So I’ve at all times been all for how humanity and know-how co-evolve over time. You recognize, we’re who we’re in the present day due to know-how.
Proper now, there’s a number of pushback towards the usage of AI in artwork. There’s a number of comprehensible unhappiness about know-how that permits you to simply press a button and get a picture. Persons are sad that these instruments have been even made and argue that the makers of those instruments, like OpenAI, ought to possibly carry some extra accountability. However right here you might be, immersed within the artwork world, persevering with to make enjoyable, participating artwork. I’m questioning what your expertise of these sorts of conversations has been?
Yeah. In order I’m certain , being within the media, the destructive voices are at all times louder. The people who find themselves utilizing these instruments in constructive methods aren’t fairly as loud typically.
However, I imply, it’s additionally a really large problem. Folks take a destructive view for a lot of completely different causes. Some folks fear in regards to the knowledge units, some folks fear about job alternative. Different folks fear about, , disinformation and the world being flooded with media. And so they’re all legitimate considerations.
After I speak about this, I am going to the historical past of images. What we’re seeing in the present day is mainly a parallel of what occurred again then. There are now not artists who paint merchandise for a residing—like, who paint cans of peaches for an commercial in {a magazine} or on a billboard. However that was once a job, proper? Images eradicated that swath of oldsters.
You recognize, you used the phrase—I wrote it down—“simply press a button and get a picture,” which additionally jogs my memory of images. Anybody can push a button and get a picture, however to be a fine-art photographer, it takes a number of ability. Simply because art work is fast to make doesn’t essentially imply it’s any worse than, like, somebody sculpting one thing for 60 years out of marble. They’re various things.
AI is transferring quick. We’ve moved previous the equal of wet-plate images utilizing cyanide. However we’re definitely not within the Polaroid part fairly but. We’re nonetheless coming to phrases with what this implies, each in a fine-art sense but in addition for jobs.
However, yeah, your query has so many sides. We might choose any considered one of them and go at it. There’s positively a number of legitimate considerations on the market. However I additionally suppose trying on the historical past of know-how, and the way it’s really empowered artists and other people to make new issues, is essential as effectively.
There’s one other line of argument that when you have a doubtlessly infinite provide of AI-generated pictures, it devalues creativity. I’m curious in regards to the steadiness you see in your work between what you do and what the know-how does for you. How do you relate that steadiness to this query of worth, and the place we discover worth in artwork?
Positive, worth in artwork—there’s an financial sense and there’s a vital sense, proper? In an financial sense, you possibly can tape a banana to a wall and promote it for 30,000 {dollars}. It’s simply who’s keen to purchase it or no matter.
In a vital sense, once more, going again to images, the world is flooded with pictures and there are nonetheless folks making nice images on the market. And there are individuals who set themselves aside by doing one thing that’s completely different.
I mess around with these concepts. Somewhat bit like—, the plunger work was the primary one. [The Plungers is an set up that Reben made by making a bodily model of an art work invented by GPT-3.] I received GPT to explain an art work that didn’t exist; then I made it. Which type of flips the concept of authorship on his head however nonetheless required me to undergo hundreds of outputs to seek out one which was humorous sufficient to make.
Again then GPT wasn’t a chatbot. I spent a superb month developing with the start bits of texts—like, wall labels subsequent to artwork in museums—and getting GPT to finish them.
I additionally actually like your ear sculpture, Ear we go once more. It’s a sculpture described by GPT-3, visualized by DALL-E, and carved out of marble by a robotic. It’s kind of like a waterfall, with one type of software program feeding the following.
When text-to-image got here out, it made apparent sense to feed it the descriptions of artworks I’d been producing. It’s a sequence, kind of forwards and backwards, human to machine again to human. That ear, specifically: it begins with an outline that’s fed into DALL-E, however then that picture was became a 3D mannequin by a human 3D artist.
And after that it was carved by robots. However the robots get solely to date with the element, so human sculptors have to come back in and end it by hand. I’ve made 10 or 15 permutations of this, taking part in with these back-and-forths, chaining know-how collectively. And the ultimate factor that occurs now could be that I’ll take an image of the art work and get GPT-Four to create the wall label for it.
Yeah, that retains developing in your work, the completely different ways in which people and machines work together.
You recognize, I made some movies of the method of this stuff being made to point out what number of artisans have been employed in making them. There are nonetheless enormous industries the place I can see AI rising work for people, individuals who will make stuff that AI comes up with.
I’m struck by the serendipity that usually comes with generative instruments, making artwork out of one thing random. Do you see a connection between your work and located artwork or ready-mades, like Duchamp’s Fountain? I imply, you’re possibly not simply coming throughout a urinal and considering, “Oh, that’s cool.” However once you mess around with these instruments, sooner or later you could get one thing introduced to you that you just react to and suppose, “I can use that.”
For certain. Yeah, it really jogs my memory a little bit bit extra of road images, which I used to do once I was in faculty in New York Metropolis, the place you’d simply type of roam round and look ahead to one thing to encourage you. Then you definitely’d set your self as much as seize the picture in the best way that you just wished. It’s type of like that for certain. There’s positively a curatorial course of to it. There’s a means of discovering issues, which I believe is attention-grabbing.
We talked about images. Images modified the artwork that got here after it. You recognize, you had actions the place folks wished to attempt to get at a actuality that wasn’t photographic actuality—issues like Impressionism, and Cubism or Picasso. Do you suppose we’ll see one thing related occurring due to AI?
I believe so. Any new inventive software positively modifications the sphere as folks work out not solely find out how to use that software however find out how to differentiate themselves from what that software can do.
Speaking of AI as a software—do you suppose that artwork will at all times be one thing made by people? That irrespective of how good the tech will get, it’s going to at all times simply be a software? You recognize, the best way you’ve strung collectively these completely different AIs—you possibly can try this with out being within the loop. You possibly can simply have some type of curator AI on the finish that chooses what it likes greatest. Would that ever be artwork?
I even have a few works through which an AI creates a picture, makes use of the picture to create a brand new picture, and simply retains going. However I believe even in a super-automated course of you may return far sufficient to seek out some human someplace who decided to do one thing. Like, possibly they selected what knowledge set to make use of.
We’d see resort rooms full of robotic work. I imply, stuff we hardly even have a look at, that by no means even makes its means by human curation.
I suppose the query is absolutely how a lot human involvement is required to make one thing artwork. Is there a threshold or, like, a share of involvement? It’s a superb query.
Yeah, I suppose it’s like, is it nonetheless artwork if there’s nobody there to see it?
You recognize, what’s and isn’t artwork is a kind of questions that has been requested without end. I believe extra to the purpose is: What is nice artwork versus dangerous artwork? And that’s very private.
However I believe people are at all times going to be doing these things. We are going to nonetheless be portray within the far future, even when robots are making work.
