Give this AI a few descriptive words and it produces an amazing picture – but is it art?

A picture may be worth a thousand words, but thanks to an artificial intelligence program called DALL-E 2you can get a professional looking image with much fewer numbers.

DALL-E 2 is new neural network An algorithm that creates an image from a short phrase or short sentence you provide. the programAnnounced by the OpenAI artificial intelligence research lab in April 2022, it has not yet been released to the public. But a small and growing number of people – myself included – have been given access to try it out.

As a researcher studying the relationship between technology and artI was eager to see how successful the program would be. After hours of experimentation, it is clear that DALL-E – although not without its shortcomings – far ahead of current image creation technology. It raises immediate questions about how these technologies are changing how art is made and consumed. It also raises questions about what it means to be creative when DALL-E 2 appears to automate much of the creative process itself.

Amazing collection of style and themes

OpenAI researchers built DALL-E 2 From a huge collection of photos with a comment. They collected some pictures online and licensed others.

Using DALL-E 2 is a lot like searching for an image on the web: you type a short phrase into a text box, and it returns six images.

But instead of excluding them from the web, the program creates six completely new images, each reflecting a specific version of the entered phrase. (Until recently, the program produced 10 images for each claim.) For example, when some friends and I were given a DALL-E 2, a “cats in devo hats” prompt, 10 photos produced That came in different styles.

Almost all of them can get professional photographs or graphics. While the algorithm did not fully understand ‘Devo hat’ – alien helmets Worn by the New Wave band Devo – headgear approached in the photos they produce.

Over the past few years, a small community of artists have used neural network algorithms to produce art. Many of these artworks have distinctive qualities that are almost similar to real pictures, But with strange distortions in space A kind of cyberpunk cub. State-of-the-art text-to-image conversion systems It often produces dreamy and imaginative images It can be fun but it rarely feels real.

DALL-E 2 offers a huge leap in the quality and realism of images. It can also simulate certain patterns with remarkable accuracy. If you want photos that look like real photos, it will produce six life-like photos. If you want Shrek’s prehistoric paintings, it will generate six portraits of Shrek as if they were drawn by a prehistoric artist.

It’s amazing that the algorithm can do this. It takes less than a minute to create each set of images. Not all images will look pleasing to the eye, and they do not necessarily reflect what is on your mind. But, even with having to sift through several outputs or try different text prompts, there’s no other current way to get so many great results out so quickly – not even by hiring an artist. And sometimes the unexpected results are the best.

In principle, Anyone with sufficient resources and experience can create a system like this. Google search Recently announced an impressive text-to-image conversion systemand an independent developer generally develops their own version Anyone can try now on the webalthough not as good as DALL-E or Google’s yet.

It’s easy to imagine that these tools are changing the way people make photos and communicate, whether via memes, greeting cards, ads — and yes art.

Where is the art in that?

I had a moment early on while using DALL-E 2 to create different types of panels, in all different styles – like “Odilon Redon Seattle Painting”—when I realized this was better than any painting algorithm I had ever developed. Then I realized it was, in a way, a better painter than me.

In fact, no human can do what DALL-E 2: generates such a diverse range of high quality images in just a few seconds. If someone told you that someone made all these pictures, of course you would say they were creative.

But THIS DOESN’T MAKE DALL-E 2 AN ARTIST. Although it sometimes looks like magic, it is still a computer algorithm under the hood, strictly following the instructions from the algorithm authors at OpenAI.

If these images succeed as art, they are a product of how the algorithm was designed, the images they were trained on, and most importantly, how the artists use them.

You might be tempted to say that there are few technical advantages in an image produced with just a few keystrokes. But from my point of view, this line of thinking resonates classic takes That photography could not be an art because the machine did all the work. Today, the human authorship and craft involved in artistic photography are acknowledged, and critics recognize that the best photography involves much more than just pressing a button.

However, we often discuss artwork as if it came directly from the artist’s intent. The artist intended to show something, or express an emotion, and so they made this picture. DALL-E 2 seems to cut this process down completely: Have an idea, write it down, and you’re done.

But when I paint the old-fashioned way, I find that my paintings come from the exploratory process, not just from the implementation of my initial goals. This is true for many artists.

Take Paul McCartney, who created the track”get back“During a jam session. He didn’t start with a song plan; he just started playing and experimenting The band developed it from there.

artist He described his process similarly: “I don’t know beforehand what I’m going to put on the canvas any more than I decide beforehand what colors I’m going to use…Every time I undertake to paint a picture I feel the sensation of jumping into space.”

in My own explorations with DALL-E 2one idea will lead to another leading to another, and eventually I find myself in a completely new and unexpected magical territory, very far from where I started.

directed shroud

I would argue that the art, using a system like DALL-E 2, comes not just from the final text prompt, but in the entire creative process that led to that prompt. Different artists will follow different processes and end up with different results that reflect their approaches, skills, and obsessions.

I began to see my experiments as a set of sequences, each statically diving into a single topic, rather than as a set of strange independent images.

Ideas for these pictures and series came from everywhere, often associated with a group of stepping stones. At some point, while making images based on the work of contemporary artists, I wanted to create a picture of the site’s installation art in the style of a contemporary Japanese artist. Yayoi Kusama. After trying some unsatisfactory websites, I came up with an idea to put them in La Mezquita, a former mosque and church in Cordoba, Spain. I sent Image to a fellow architect, Manuel Ladron de Guevara, from Córdoba, and we began to draw on other architectural ideas together.

This has become a series of imaginative new buildings in different architect styles.

So I began to think of what I do with the DALL-E 2 as a form of exploration as well as a form of art, even if it’s often amateur art like the graphics I do on my iPad.

In fact, some artists like Ryan MurdochShe advocated that instant photography be recognized as an art. pointing to Experienced AI artist Helena Sarin as an example.

“When I look at most things from MidjourneyAnother popular text-to-image system — “a lot of it would be fun or interesting,” Murdoch told me in an interview. but with [Sarin’s] Work, there is a line. It is easy to see that she has thought a lot about it, and worked on the craft, because the output is more visually appealing and interesting, and it follows her style in a continuous way.”

Working with DALL-E 2, or any of the new text-to-image systems, means learning the quirks and developing strategies to avoid common pitfalls. It is also important to know more Possible damage, such as its reliance on stereotypes and potential uses for misinformation. With DALL-E 2, you’ll also discover surprising associations, like the way everything gets old when you use the style of an old painter, filmmaker, or photographer.

When I have something very specific that I want to make, DALL-E 2 often can’t do it. The results will require a lot of difficult manual editing afterwards. When my goals are vague, the process is a lot of fun, offering surprises that lead to new ideas that in turn lead to more ideas and so on.

Formulating new facts

Text-to-image systems can help users imagine new possibilities as well.

Activist actress Danielle Baskin She tells me that she always works “to show alternate realities through the ‘real’ example: either by setting up scenarios in the physical world or doing exact work in Photoshop.” However, DALL-E 2 “is an amazing acronym because it’s so good at realism. And that’s the key to helping others bring to life a potential future — whether it’s satire, dreams, or beauty.”

I used it to imagine alternative transmission system And the Plumbing that moves pasta instead of waterboth of which reflect Her sensitivity is a provocative artist.

Likewise, artist Mario Klingman Architectural designs with tents for the homeless can be taken as a response to Architectural perceptions of luxury dream homes.

It is too early to judge the significance of this art form. I can still think of a phrase from the excellent book.”Art in Beyond Culture“-” Prevalent in artificial intelligence is modernity. “

This would certainly be true, to some extent, of any new technology used in art. first movies Lumiere brothers In the 1890s there were creations, not cinematic masterpieces. People were amazed to see the pictures move at all.

The AI ​​art program is evolving so fast that there is constant technical and artistic novelty. It seems that, every year, there is an opportunity to explore a new and exciting technology – each one more powerful than the previous, and each seems poised to transform art and society.