Alex Reben's works are often absurd and surreal. He created a hybrid image of a giant ear using DALL-E, which was then hand-carved by humans from marble.

The message conveyed by this piece is relevant to everyone. Reben shows a great interest in the role of humans in a world full of machines and how these roles are changing.

Reben said, "I use humor and absurdity to deal with many such issues. Some artists may confront issues in a very serious way, but I find that if you add a touch of absurdity, it makes the ideas more acceptable, even if you are trying to tell a very serious story."

Reben is the first-ever artist-in-residence at OpenAI. Strictly speaking, this appointment began in January 2024 and lasted for three months.

Advertisement

However, Reben's relationship with OpenAI seems casual: "This work is a bit ambiguous because I am the first, and we are all feeling our way forward. I may continue to collaborate with them."In fact, Reben has been collaborating with OpenAI for many years. Five years ago, before the public release of GPT-3, he was invited to try its early versions.

He said: "I tried a lot of things and made some artworks. OpenAI was very interested to see how I used their system in different ways. I agreed because I also wanted to try something new. At that time, I was mainly making things with my own models, or using websites like Ganbreeder (the predecessor of generative image-making models)."

In 2008, Reben was studying mathematics and robotics at the MIT Media Lab. There, he helped create a cardboard robot named Boxie, from which the lovable robot Baymax (Big White) in the movie "Big Hero 6" drew inspiration.

He is now the Director of Technology and Research at Stochastic Labs, a non-profit artist and engineer incubator in Berkeley, California, USA.

I chatted with Reben via Zoom about his work, the unresolved tension between art and technology, and the future of human creativity.Here is the translation of the provided text into English:

Here is our conversation, with the content edited.

MIT Technology Review: You are interested in the way humans interact with machines. As an artificial intelligence artist, how would you describe your use of technology? Is it a tool or a collaborator?

Reben: First of all, I do not call myself an artificial intelligence artist. Artificial intelligence is just another technological tool. If something else comes along after artificial intelligence that I am interested in, I will not say, "Oh, I am just an artificial intelligence artist."

MIT Technology Review: Okay, but what about these artificial intelligence tools? Why has your career been playing with this technology?

Reben: My research at the MIT Media Lab is all about social robots, studying how people and robots can be combined in different ways.

(Note: The original text seems to be cut off, so the translation ends where the text ends.)The robot Boxie is also a filmmaker, capable of interviewing humans. We have found that people are willing to open up to robots and tell them very profound stories. This was before the popularization of artificial intelligence virtual assistants like Siri.

Nowadays, people are already accustomed to conversing with machines. Therefore, I have always been interested in how humans and technology have co-evolved over time. You know, we are who we are today because of technology.

Image | A cardboard robot and a plastic mask artwork titled "Five Dollars Can Save the Earth" (Source: COURTESY OF ALEXANDER REBEN)

MIT Technology Review: Nowadays, there are many people who oppose the use of artificial intelligence in art. There is a lot of dissatisfaction with technology that can produce images with just the press of a button.

People are even dissatisfied with the manufacturing of these tools and believe that developers of these tools, such as OpenAI, should take more responsibility. But you are still immersed in the art world, continuing to create fun and immerse yourself in art. I would like to know what you think of these voices?Leiben: Yes, you know, in the media, the negative voices are always louder. Those who use these tools in a positive way are sometimes not as loud.

MIT Technology Review: But, I mean, this is also a very broad issue. People have negative views for many different reasons.

Some people worry about the datasets, some worry about job loss. Others worry about misinformation and the world being flooded with media created by artificial intelligence. These are all legitimate concerns.

Leiben: When I talk about this, I will take photography as an example. What we see today is basically similar to what happened at that time.

There are no longer artists who make a living by painting for goods, such as painting peach cans for advertisements in magazines or on billboards. But that was once a job, and photography has eliminated that group of people.You just said a sentence, I wrote it down, "Just press a button and you can get a photo," which also reminded me of photography.

Anyone can press the button and take a photo, but to become an excellent artistic photographer, it takes a lot of skills. The fact that an art piece is made quickly does not necessarily mean it is worse than something carved in marble for 60 years. They are different things.

MIT Technology Review: Artificial intelligence is developing rapidly. Analogous to the field of photography, it's as if we have already surpassed the wet collodion photography technique using cyanide, but we certainly have not yet entered the Polaroid stage. We are still thinking about what this means, both in terms of artistic significance and employment.

Reben: Yes, but your question covers many aspects. We can choose any of them and then do it.

There are definitely many reasonable concerns. But I also think it's important to examine the history of technology and how it has truly empowered artists and people to create new things.MIT Technology Review: There is another perspective that suggests if you have an infinite number of AI-generated images, the value of creativity is diminished.

I'm curious, what kind of balance have you seen in your work: what you do versus what technology does for you. How do you connect this balance to the issue of value, and what value can we find in art?

Reben: The value of art has an economic aspect and a critical aspect, right? From an economic standpoint, you can tie a banana to a wall and sell it for $30,000. It purely depends on who is willing to buy it.

From a critical standpoint, going back to the example of photography, there are countless images and photos in the world, but there are still people who continue to take great photographs. There are also those who stand out by doing something unconventional.

I try out many different ideas. Plunger was the first work (Editor's note: Plunger is a real work created by Reben according to the invention of GPT-3). I had GPT describe a non-existent artwork, and then I made it.This slightly subverted my definition of the author, but (the creative process) still requires me to browse thousands of outputs to find one that is interesting enough.

At that time, GPT was not a chatbot. I spent a month writing some texts, such as wall labels next to museum art pieces, and let GPT complete them.

MIT Technology Review: I also like your ear sculpture very much. This is a sculpture described by GPT-3, visualized by DALL-E, and finally carved by a robot in marble. It's a bit like a "waterfall development process," and it's one software after another.

Reben: When the text-to-image model appeared, it made sense to provide it with descriptions of the artworks I had been generating.

This is a chain, with back and forth, from human to machine, and from machine to human. Especially that ear work: it started with a description input into DALL-E, but later this image was turned into a 3D model by a human 3D artist.Afterward, it was carved out by a robot. However, the robot can only complete a limited amount of detail, so a human sculptor must intervene and finish it by hand. I have tried similar work 10 to 15 times, back and forth, linking the technology together.

Now the last thing in the chain is that I will take a photo of the artwork and let GPT-4 create a wall label for it.

MIT Technology Review: This constantly appears in your work, the different ways of interaction between humans and machines.

Leiben: I have made some videos showing the production process of these things to show how many craftsmen are employed to make them.

In some other large industries, I can still see that artificial intelligence has increased people's work, and they will manufacture what artificial intelligence can produce.MIT Technology Review: I am struck by the serendipitous discoveries that generative tools bring, creating art in a world full of randomness. Do you think there is a connection between your work and existing art or readymades, such as Duchamp's "Fountain"?

My point is, you might not see a urinal and think it's "too cool." But as you play with these tools, at some point, you will find some highlights and react to them, thinking about how you should make use of it.

Reben: Of course, yes. It actually reminds me more of street photography, which I often did when I was in college in New York, where you would wander around, waiting for something to inspire you.

Then you can take the photo in the way you want. The two are somewhat similar. It definitely has a curating process. It's a process of looking for things, and I think that's interesting.

MIT Technology Review: We discussed photography. Photography has changed art after it. Some movements emerged where people wanted to try to capture a "reality that cameras can't capture," such as Impressionism, Cubism, or Picasso. Do you think we will see similar things happen because of artificial intelligence?Leiben: I suppose so. Any new artistic tool will undoubtedly change the field, as people will not only discover how to use the tool but also how to distinguish themselves from what the tool can achieve.

MIT Technology Review: Speaking of using artificial intelligence as a tool, do you think art will always be made by humans? No matter how good the technology is, it is always just a tool?

By stringing these different artificial intelligences together, you can do this without any participation at all. You could have some kind of curatorial artificial intelligence that selects its favorite things. Would that be art?

Leiben: In fact, I have several works where artificial intelligence created an image, used that image to create a new image, and kept the cycle going.

But I think that even in a super-automated process, you can trace back to a far enough stage where you find some people making decisions. For example, perhaps they chose the dataset to be used, which is one of the decisions.We might see hotel rooms filled with paintings by robots. What I mean is, things we hardly care about, or even don't need people to plan at all.

I think the real question is, how many people need to be involved in making a piece of art. Is there a threshold, or a percentage of participation? That's a good question.

MIT Technology Review: Yes, I think it's like, if no one is looking at it, is it still art?

Reben: What is art and what is not art is one of the questions people keep asking. I think what's more important is: what is good art and what is bad art? That's very personal.

But I think humans will always do these things. In the distant future, even if robots are making images, we will still paint.