AI art. Putting the art in artificial intelligence

Teddy bears working on new AI research underwater with 1990s tech. An astronaut playing with cats as pixel art. A bowl of soup that is a portal to another dimension drawn on a cave wall.

These are just a few of the infinite bizarre combinations of prompts that users can type into DALL-E 2, an open-source AI platform, which lets users generate art pieces based on textual prompts that they type in. What started as a research project is now available to users as a beta version: with a promise of photorealism, great resolution, and accurate images.

Generating AI art on Dream by Wombo

AI art generated on Dream with the prompt ‘A pond filled with blooming lilies and some frogs’

Generative AI is now everywhere- beyond art, OpenAI’s ChatGPT mimics human conversations and can write things from student essays to computer programmes. There is also Lensa AI, an AI-based photo editing app that allows users to generate avatars from pre-existing photos. From code to conversations, art to music compositions, anything seems possible to be generated, by anyone, with artificial intelligence at the helm.

Also read: Lensa AI: Is this app safe? 

As this wave surges across online platforms, the talk surrounding AI art is growing louder, with digital artists voicing their views and critique on AI art.

Learning curve

Mira Malhotra, Founder of Studio Kohl, a graphic design studio, detailed her experience with different AI text-to-image websites. “Midjourney was a bit of a learning curve because I had to first learn how to use Discord, which wasn’t very user-friendly. DALL-E was easier to use but generated a lot of “cursed” faces and immediately assumed I needed close crops. Wonder AI had similar issues,” she said.

“Since then, I’ve studied how prompts are built to generate pleasing and usable images, and I see that while great flukes are definitely possible, users need to be very specific on how they write prompts, and for that, you definitely do have to have experience in art.” Malhotra also said that “AI is creating a very cookie-cutter, banal, and uninspiring bunch of images unless used correctly by a real creative.”

The criticism

AI art is, by no means, free of criticism. Some artists complain about the ‘Uncanny valley’ effect of AI-generated faces, while some stress the need for proper regulation. Regardless, the widespread backlash against AI art has swept art forums, with “No AI Art” images flooding sites, such as ArtStation and Twitter. The online discourse began when artists such as @ZakugaMignon tweeted the image, writing, “Ai “art” is currently scraping the web for art and uses it in datasets. No artist gave consent to have their art used. We were not compensated.”

The need for clear differentiation then becomes pertinent. Jayesh Joshi, art director of Schbang, a creative agency, believes that AI art should be treated differently from digital art. “Man-made and machine-made art are clumped together, which isn’t fair.”

A paramount concern that many digital artists have expressed is that several AI art generators use an image bank of pre-existing, man-made art with no scope of crediting the original artist. Joshi further said, “AI development companies should be mindful of what they feed this image bank, and use their platform to uplift artists whose works they’re using - possibly commissioning the creators.”

Striking a balance

Another artist, Aditya Mehta, CEO of Art&Found, observed the need to strike a balance between AI and man-made art. “It is interesting to see how AI and Machine Learning keep evolving, which is an inevitable and endless pursuit of how AI can co-exist with and help humanity.

My view is that you can’t compare the two. You can have them co-exist and collaborate. Art simply created using AI-generated art tools doesn’t excite or shock me. What excites me is seeing how someone uses them to tell a compelling story,” he said.

Also read: As AI rises, lawmakers try to catch up

To reach such a level of collaboration, AI art can be deployed for several purposes, including the generation of reference images that allows artists to input their vision through prompts. The nuance here, according to these artists, is that AI art is used as an aiding tool, not the medium itself.

“AI art can also streamline processes and help in developing quick thumbnail sketches, palette experiments, composition layouts, and lighting variations. If done right, AI can assist human artists in creating brilliant new work,” added Joshi.

With the spotlight on AI art shining brighter, one thing seems certain: it is going to be near impossible to nip its growth in the bud. “I see the spread of AI growing exponentially in 2023 to memes, metaverse, animation, films, CGI, gaming, comics, graphic novels, communication aid, and education,” said Mehta. “And one Black Mirror episode.”

COMMENT NOW