Throughout 2022, the rise of AI has been plain to see. This is most plainly seen within memes which have made their rounds on places such as Reddit and Twitter. One Twitter ai bot generator, which showcases some of the funniest and oddest ai art generations, currently has 1.2 million followers. Many people have taken to the internet to post silly images based on prompts from AI art generators.
In June, Cosmopolitan revealed its first AI art magazine cover, doting how it ‘only took 20 seconds to make’ despite having ‘shockingly good’ results.
Cosmopolitan presenting the first AI-generated magazine cover.
But how does AI art get made exactly? How does it take 20 seconds to generate an image that might have taken a human artist hours or a photographer hundreds of shots to get right?
AI-driven art tools are curated through learning. AI has no form of creativity, so it needs to receive human input to operate. These numerous AI art tools have to be fed large databanks of human-produced photographs and artwork pieces. So that the AI can reassemble them as new images based on prompts received through human input, its generations are rule-based and developed through mathematics, which is very different to how a human artist might potentially conceive a piece of artwork. The most popular AI art generation model as of 2022 is called ‘Stable Diffusion’.
Many AI image creators are also often either larger corporations or groups of individuals who do not produce art themselves. So, art generated by many of these programs has needed to source art from external means. Most often, taking art produced by other artists and posted on the internet across almost all social media sites for the initial generations to be able to make images. This is done without the artist’s permission in most cases. Such an act can easily be considered art theft.
Once these AI tend to go public, it can use people’s prompts to further ‘learn’ what images to generate. Alternatively, it can receive inputs from user-submitted images (such as in the case of Different Dimension Me and other AI art programs that make interpretations based on other images.) However, this still means that the usage of other artists’ artwork without permission is still required to create the algorithm.
Traditional and digital artists have been protesting the recent upsurge in AI art and its usage. More recently, users on the site Art Station have started protesting after AI-generated images were revealed to be allowed on the site.
This image has been mass posted to Artstation’s website as a form of protest.
Another art curation website, Deviantart, also recently revealed that users who uploaded artwork would be used to train an AI art generator of their own making by default.
Other issues with AI art that traditional and digital artists have pointed out include creating a culture of distrust between artist and client. As well AI provides unfair competition between human artists and AI algorithms.
There has also been fear among artists and the general public that AI art might even replace human artists completely. Or, at the very least, leave many human artists unemployed.
Another issue with AI-generated art goes beyond that of human artists and their employment, which is the behaviour of many companies behind these new AI art software.
The AI anime art Different Dimension Me is owned by the Chinese company Tencent. A company which has been known to be in a close relationship with the Chinese Government, both receiving funds from it and aiding it with its social movements. In one example, Tencent collaborated with the Chinese Communist Party and the People’s Daily to develop “patriotic games.” One can only wonder what Tencent might do with AI art tools.
In one example, the AI art program produced by Tencent has also been known to ‘whitewash’ various dark-skinned people as it attempts to translate real-life photos into an anime style.
Twitter user @fyridk gave the app an image of the Jackson 5, resulting in an image that looked nothing like the Jackson 5—converting them into very light-skinned anime girls with hats.
Twitter user @fyridk’s post regarding the Jackson 5.
It’s clear that despite AI apps learning from the data presented to them, an awful lot of AI programs, both art AI and other forms, is that it is prone to biases. Due to a mixture of lack of training and humans being prone to natural biases, including unconscious bias, which will often be missed by programmers.
Other concerns raised include the fear of where images used by these AI programs might go. Because Tencent and other major corporations are often the ones behind creating these tools, people fear that their own personal images or prompts that they have provided might be used to power government surveillance tools or another tracking AI which corporations might utilise against their workforce. Even though a simple AI might be primarily dedicated to the production of artwork, the information it generates may be used for ulterior motives by some groups, especially powerful ones.
With its boom across the year, AI-generated art and the debate surrounding it will not likely go away soon. Whether society will keep AI art for relatively light-hearted memes or for replacing art industry giants as a whole is yet to be seen.
Share This Post On
0 comments
Leave a comment
You need to login to leave a comment. Log-in
