When AI takes on the role of an artist

Artificial intelligence can now create any image in seconds from a text prompt.Image: Shutterstock

AhArtificial intelligence is increasingly carving out a place for itself in the visual arts. The software can now generate images from a simple sentence or series of prompts. The results are amazing, but using such a program raises many questions about the future of creation and creativity. We take a look at this phenomenon in our latest After Calendar, our trend book for 2023.

In 1901, American architect Frank Lloyd Wright gave a lecture in Chicago on the role of machines in the creative process. He explained: [was] Finally [to] ‘Liberating Human Expression’ A century later, this ‘machine’ takes the form of artificial intelligence (AI). Create any image from a text description (text prompt) in seconds. Would you like to see British naturalist Sir David Attenborough fight a polar bear with his bare hands? Or American-Canadian actor Dwayne Johnson meets his alter ego, The Rock? Thanks to software like Midjourney, DALL-E 2, Imagen, DreamBooth and Stable Diffusion, it’s possible.

These tools use language understanding and learning models on vast amounts of data to generate images from lines of text. Google explains that it trained Imagen on his LAION-400M. LAION-400M is a database of 400 million images associated with written captions found on the Internet. This is where artificial intelligence gets its “inspiration”. It then generates several illustrations corresponding to the request or prompt (eg, a teapot with muscles) by reducing image noise (denoising) in clusters of randomly colored pixels. It is this “diffusion” technique that makes modern image generation software so successful.

Since June, we’ve seen a surge in popularity on social networks such as DALL-E 2, Midjourney, and others. This is mainly due to the realistic and grotesque visuals they produce. Some present fairly vague concepts like blockchain and joy, while others address very specific requests (such as “a quail listening to an angry parent at a school district board meeting”). ). Generative artificial intelligence software can respond (more or less gracefully) to any request submitted by a user as long as it does not violate moderation policies. For example, OpenAI, the company behind the DALL-E program, bans the production of violent, pornographic, or political imagery, specifically preventing the creation of “photorealistic” illustrations of celebrities. I’m here. DALL-E therefore refuses to imagine a battle between Donald Trump and Vladimir Putin.

Also read: Subodh Gupta’s giant paraphernalia sweeps glamorous Parisian stores

imitation and limitation

But, in Frank Lloyd Wright’s words, this is pretty much the only limit this technology imposes on “human expression.” Programs such as Imagen and Midjourney don’t just make image compositing available on his web. They also create new ones. Draw in the style of Picasso, Pop Art, Impressionism, Street Art, or even Cyberpunk or Bauhaus. takes some time. As you might think, just entering random keywords as a prompt isn’t enough to get a unanimously high rating. Digital artist Erik Carter told his MIT Technology Review:

In particular, these programs have difficulty imagining anatomically correct animals and humans. Most of the faces they produce look “squashed” or at least distorted. Philosopher and journalist Tom Whyman wrote in ArtReview that the AI-designed characters “appear exactly as imagined, but not quite.” Another concern he has is that generative AI tools reproduce certain racial and gender stereotypes. DALL-E 2, Midjourney, etc. tend to represent caregivers as women and doctors as men. It also conveys very western normative images due to the lack of diversity in the millions of images it uses for training.

OpenAI has already announced various measures to prevent DALL-E 2 from becoming a vehicle for further discrimination on the internet. The company, like the company, is endorsed by millions of Internet users who want to improve the images generated by their flagship program. Hundreds of thousands of users flock to his DALL-E 2 Discord community server to share tips on how to improve text descriptions (a process called “prompt crafting”) and how to overcome the limitations of artificial intelligence. are exchanging. “Do you have any advice for generating characters for popular TV shows in Dall-E 2? It’s fine to come up with heroes like SpongeBob SquarePants, but for Patrick Star, for example, I’d rather have a starfish than the character himself.” ,” asks one user. He has already had one member of this online community help him create a centaur.

This phenomenon is not limited to DALL-E 2 or Discord. Users of generative artificial intelligence software are also gathering with other enthusiasts on Reddit and Twitter to discuss possibilities. Some enthusiasts, like Guy Parsons, publish digital guides to teach readers how to best use these new art-making tools.

Also read: Truly tasteless: Japanese plastic food artist gets creative

“Harmful” Technology?

Last August, American artist Jason Allen won an art contest held inside the Colorado State Fair with Midjourney’s artificial intelligence-generated painting. Entitled ‘Theater de Opera Spacial’ or ‘Space Opera Theater’, the work in question required 80 hours of work, or more precisely 80 hours of ‘rapid production’. However, when the award was announced, it caused controversy in the art world, albeit modestly. Some creators saw it as a sign, as one Twitter user suggested, that they were “watching the death of artistry unfold before their very eyes.” Some fear that the rise of the program will contribute to the “Uberization” of the creative industry.

The terror caused by the San Francisco Ballet (SFB)’s recent “Nutcracker” campaign. An American dance company used an illustration created by Midjourney to promote a famous ballet, but many internet users of his were unimpressed. Some turned to social networks to criticize the SFB’s decision to use an AI image generator instead of the illustrator of “Flesh and Blood”. It is very disappointing to see you actively harming an artist,” one user wrote below SFB’s Instagram post. “AI art in its current form uses the artist’s work without consent, credit, or compensation.”

Faced with growing interest in this AI image-generating program, some artists are stepping up their resistance. His one place where they have voiced their displeasure is the ArtStation platform where many of them host their work. It all started when he noticed in December that images created by artificial intelligence software were being uploaded to the platform. In the view of Ghostfire Gaming’s art her director, Suzanne Helmigh, this is unacceptable. “I’m really considering removing my art online on places like Artstation and Deviant Art,” she said on Twitter. If only to be sent, exploited, and Frankensteined into AI visuals.”

Images were uploaded to ArtStation protesting and denouncing the existence of these controversial works alongside the artist’s work, using the hashtag #noaiart in support of protests by digital artists using the platform. ArtStation responded to the controversy with an AI FAQ containing the statement that “ArtStation’s Content Guidelines do not prohibit the use of AI in the process of submitting artwork.” The Epic Games subsidiary “doesn’t want to be the gatekeeper of site conditions that stifle AI research and commercialization,” adding that artists should be free to decide how their work is used. emphasized. He sided with ArtStation by welcoming AI-generated images with open arms, while some image banks like Shutterstock ban them.

tools rather than artists

And what about the place of museums and other cultural institutions in all of this? Ready to host images generated by DALL-E 2, Midjourney and more, along with works by Leonardo da Vinci, Pablo Picasso and Yayoi Kusama? Are you? This idea is still under discussion as it involves ethical issues regarding the concept of these images, transparency about the process, and consent of the authors of the works used to train the artificial intelligence software. In addition, questions arise regarding the role of machines as actors or as passive technical elements. Is it something that controls the creation, or is it just another tool for the visual his artist, unlike the camera?

The Museum of Modern Art in New York believes this technology offers an opportunity to showcase its uniquely rich collection in new ways. Turkish-American Digital He invited Refik Anadol, his artist, to use the museum’s collection of nearly 200,000 works of art to explain the history of contemporary art with the help of artificial intelligence. I’m here. Visitors to the New York Museum can see the final result through several immersive installations collectively called “Refik Anadol: Unsupervised” until March 5. As Refik Anadolu explains, these installations take the form of his three protean data sculptures showing machine ‘dreams’ and ‘hallucinations’.

According to Michelle Kuo, MoMa curator and co-organizer of the exhibition Refik Anadol: Unsupervised, ever-changing dreams reflect art. “This project reshapes the relationship between physical and virtual, real and unreal,” she said in a statement. “AI is often used to classify, process and generate realistic representations of the world. In contrast, Anadolu’s work is prescient.It explores dreams, hallucinations and the irrational. , presents a different understanding of contemporary art and art-making itself.”

But as aesthetically appealing as they may be, the “dreams” of machines and AI are fed by the dreams of real human creators. Even as artificial intelligence pushes the boundaries of creative human expression ever more, we still need commands and prompts from a real person to create. It lacks the intention, drive, or motivation to create the artwork. It is we humans who interpret its realization that way. Such technology is not ready to replace human artists, at least not at this time.

After Calendar:

In 2022, the words ‘better, better, less’ take on more meaning. And this isn’t necessarily all bad news. Because drinking can be accompanied by mindfulness, creativity, and even unexpected pleasures. That’s why trends for 2023 examine the joy of quitting something. A new lazy aesthetic. When AI takes over the realm of art. Yeast as an increasing ingredient. The power of women over 50. And a future where we are all energy producers. And of course, his 23-word glossary of terms to use in 2023…Find all the trends in the 2023 Aftercalendar….enjoy reading.