Artists worry that AI-made art could make it harder to sell their work or get noticed

by AZALEA AZUAR

HEATED discussions transpired in the art world after an artwork generated from artificial intelligence (AI) won an art competition. Jason M Allen received a backlash from other artists when his submission entitled “Théâtre D’opéra Spatial” won the award for emerging artists in the Colorado State Fair fine arts competition.

He used Discord-based AI art generator Midjourney and later upscaled it using AI Gigapixel to put it on canvas.

The contestant defended his submission and even suggested to the organisers to open a category for AI-generated artworks next year. Some argued that although Allen had to key in prompts to generate the artwork, he fine-tuned it for many weeks and used Photoshop to edit his work.

They opined that Midjourney is just another digital tool like Adobe Photoshop or Illustrator and the artwork is considered a digital art.

Meanwhile, others said it was unfair because many other artists took a lot of effort in creating their artworks and felt that it could lead to the death of artistry.

This then sparked another discussion of whether artists who use AI can even be considered artists.

How AI-generated Images are Created

Digital art can be created using many ways and different software. Although the process is not the same as traditional art, specific skills and knowhow are also required in order to produce good digital art.

On the other hand, AI artwork is generated through machine learning process, where AI models are fed with billions of images which they then can use as basis to create new images.

Users then provide them with text descriptions known as “messages”, and the software generates a picture that matches that description based on its training data.

The more detailed the prompts given, the better the artwork that can be generated.

AI is able to quickly generate professional-looking art — roughly in 10 seconds.

These systems have already been the exclusive domain of well-funded technology companies and have become increasingly available over the last few months.

Midjourney is one of the most popular platforms and known for its meticulous aesthetic, which often mimics contemporary digital art styles.

There is also OpenAI’s Dall-E 2 which is known to be intuitive and is able to generate artwork accurately through the prompts provided.

Users can also tweak the image if they are not satisfied with the results.

Furthermore, there are image manipulation mobile apps such as Wombo which enable users to create art with the simple entrance of a quick prompt. Users can choose different art styles and Wombo also features a free mobile desktop version.

Copyright Issues

AI models cannot create images out of the thin air and have to be trained using billions of images that are available online.

However, not all of the images used are freely available on the public domain — many are copyrighted images that belong to other individuals and organisations.

Thus, copyright has become a huge issue since some of images fed to  the  AI  training models are not necessarily meant for commercial use.

There is also no one stopping users from copyrighting AI-generated art, so in the end they are able to mass-produce and sell these generated artworks at fan conventions and online.

Another issue of AI-generated images is that the AI systems are unable to completely filter out images that may come up as offensive and harmful, although they are programmed to filter out negative representations.

Architecture graduate-turned-freelance illustrator Hazimah Harith said AI would have a negative effect on artists if big corporations take their artworks and feed the artworks into their AI systems. “It is (a problem) when they unethically take all the artworks of certain artists and use them without the artists’ permission to generate new art.

“Actually, it is not really their hard work because they cannot really beat real art,” she said to The Malaysian Reserve (TMR).

Although Hazimah confessed that AI-generated artworks are impressive, they do not have the “human touch”.

She added that if the art market is oversaturated with AI art, art itself may lose its value and people will lose their appreciation for art. “Some people cannot identify

AI art. They will be impressed by it when actually it is just AI-generated,” she said.

However, Hazimah believed that traditional art is still here to stay as some people prefer to see the real strokes unlike AI-generated artwork which are all smooth and blended.

If used well, AI can help artists in terms of productivity where it can automate mundane tasks, create visuals for marketing and help to streamline the creation process.

This way, it can help artists focus on the creative aspects of producing their artworks.

AI is also able to aid artists produce sophisticated projects where digital assets can turn artworks into something more life-like, adding elements and also identifying user trends and preferences.

AI Changing the Art Market 

According to multidisciplinary visual artist Adib Zahran who goes by the name of “Zhonk Vision”, AI is already changing the art market through innovations and understanding trends.

“Some people worry that AI-made art could make it harder for human artists to sell their work or get noticed.

“We do not know how much AI will change the art market in the future,” he told TMR.

For instance, AI might shift work cultures in creative software companies such as Adobe, like how it managed to revolutionise an independent animation studio Corridor Crew by changing its process of making short films.

“It is like a very smart robot that can use the information it has been given to make decisions on its own. “This can be helpful because it can help us figure out how to solve hard problems and make things run more smoothly,” he added.

However, Adib advised users to use AI ethically and responsibly, as well as to not heavily rely on it as it is still able to make wrong and unfair decisions.

Adib mainly works on motion and AI-generated images to transmit visions of exploration through a sequence of messages that visualise dream and fantasy.

He has a background in technology which was enhanced by his self-taught coding skills.

However, he has dabbled in the art field and involved himself in arts-related events with a marriage of technological innovation such as Hausboom Festival 2022, Hujan Rumah Kita 2022, Kuantam and Immersio 2022, as well as previously working in creative agencies.

Although Adib enjoyed creating artworks using traditional media, he preferred to showcase his works on new media.

“It is debatable whether any of this qualifies as ‘art’, but in my opinion, creative efforts should be pursued because they bring you joy and allow you to express yourself freely,” he said.

It was only in 2017 that he started to experiment with AI after working with Reka for its autonomous vehicle project.

Later, he began to use it more frequently to enhance the details and resolution for his final product. It was a struggle for Adib to choose the correct tools until he found VQGAN + Clip, PYTTI, Diffusion, Laion and GPT3, and understood the technology clearly.

“AI tools are trending to the point where you can find so many tools easily over the Internet, either from the consumer’s perspective or the developer’s point of view, especially in GitHub and Hugging Face. But getting the aesthetics and complimenting the originality is hard.

“That is why I am still learning and working with multiple AI models, aiming to utilise them as a tool for my work,” he added.

Adib aims to explore AI in creative production such as musicals, performance art, theatres and even short films as not everyone possesses special machines and expensive equipment.

“I am on the lookout and actively seeking for opportunities to collaborate with artists and filmmakers in this medium,” he said.

---

• This article first appeared in The Malaysian Reserve weekly print edition