Why Hire Fashion Models When You Can Use AI?
The month of May has been embellished with every large technology company presenting to the public a new toy involving AI, making us the guinea pigs and them the observers. First seen as a striking innovation for mankind, it is now considered a slight to artists and other creatives who are offended by AI.
To put it simply, Vogue Italia has removed the need for creative input going forward. Vogue Italia’s May 2023 Issue is the first Vogue cover to use artificial intelligence. The magazine used DALL-E in what seems to be a push for AI to be used in conjunction with photographers and stylists.
In the latest Italia issue, fashion photographer Carlijn Jacobs and visual artist Chad Nelson worked together alongside supermodel sensation Bella Hadid to create a Vogue edition filled with AI images. Chad Nelson gained a lot of press when–in only a week–, he created a short film made just with artificial intelligence photos. Though the film itself was only five minutes, the animation process took up to six months to create. But with DALL-E, an image-learning model made by San Francisco AI lab OpenAI, the process to make short films could be much faster and cheaper. DALL-E creates AI prompt-based images quickly and at a relatively low price. Though Nelson still needed a script, voice actors, and animators to bring the images to life, the overall staff to make an animation greatly decreased. In fact, a recent Goldman Sachs report said that AI could disrupt the global economy and subject 300 million jobs to automation (cnn.com).
So when Vogue is implementing the use of AI in something as influential and prevalent as their cover, suspicions begin to rise on the messaging behind it. Is Vogue trying to be radical, or is it mocking the doomsday for creativity, indifferent to the millions of jobs that will be lost? To critics, most choose the latter.
The real culprit is DALL-E. DALL-E 2, its successor which holds higher image quality and more range, is an API, meaning that developers can implement DALL-E into their applications. From this, all applications could implement this AI despite never being given the source code. Because the code is not being given to inspect, issues over bias and plagiarism arise, questioning how unique an AI image can be.
For example, the cover of Vogue Italia looks eerily similar to 35-year-old Elizaveta Porodina’s work, a Russian photographer seen in countless papers such as GQ magazine, Vanity Fair, and Vogue. Her images are surreal and colorful, reminiscent of positive dreams you can’t quite remember. Color, movement, and glamor painted across the images create memorable shots that have never been seen before.
The similarities pose questions if images by AI should be considered art if that art is simply using old photos to collage and blend “new” ones. AI can only make superficial blends of existing artwork, which poses the question of whether art is truly original. AI-generated art can even be considered theft, if these algorithms take concepts without the owner's consent.
What we do know is that no matter how hard AI tries, it will never have one thing that humans have: connection. Even if it may seem like AI-generated images are complex visually and emotionally, they will never have the same impact on the viewer as human artists. This is why all digital art seems bland, even if it is interesting to look at. This imitation of true art, human art, leaves a ghost of an image with a soul unable to be found.
However, plagiarism is still prevalent in the art world. Take most pop culture art, for example, where collages full of art made by other artists are attributed to the artist who copied them. Andy Warhol, Mr. Brainwash, and Banksy have all been accused of outright plagiarism for their art, which sells for millions. If the concept of copying art is allowed in the art world, shouldn’t AI be allowed to do this as well?
The issue with AI art is that it cannot be proven if the art is stolen. Like figuring out if a song was copied from another because of a similar tune, it is hard to assume that an AI image is taken from an artist unless it is an exact copy. Pop art, on the other hand, is visually very similar, so the debate is less disruptive.
This is not the way that AI can get into the good graces of artists. Though difficult, AI art should be so unique that there is no doubt of its authenticity. Carlijn Jacobs and Chad Nelson during the Vogue Italia experiment discussed their thoughts on playing with DALL-E. “In order to effectively communicate my vision to the AI, I had to learn how to highlight my creative ideas and concepts in my mind and translate them into specific keywords, describing my vision in great detail. I initially tried random inputs, but soon realized that precision in my instructions was crucial,” says Carlijn Jacobs.
Stylist Imruh Asha chooses to see the benefits of AI in the creative process. “On the opening page, the JW Anderson dress already had keyboard elements on it. What's fascinating is that the AI recognized this and extended the design on the right side, generating the little cables and yellow design. It's remarkable how the AI not only understands the direction of light but also interprets information from the loaded picture's pixels. The dress was seamlessly integrated with the AI-generated wires and design, showcasing how AI can enhance and extend the creative possibilities of an image,” says Asha.
If anything, AI adds to the creativity already present. However, what people should not assume is that AI is creativity in itself, or that it should be used to think of unique and creative ideas. If this becomes the norm for other professions, everyone's ideas will soon lack variety, making our world somewhat dystopian. Our thoughts would bear more similar to an algorithm than our peers. Lastly, it wouldn’t be a conversation about AI without discussing the blatant bias that AI has. Because the code is not available, the true reason for this bias may only be hypothesized, but the results can be quite daunting.
This photo in Vogue Italia immediately rubbed people the wrong way. The first observation to many was the distorted faces of the people next to model Bella Hadid, while her visage is made to perfection. Her extravagant clothing and enlarged body are juxtaposed with the people around her with presumably blue-collar jobs. Those “people” in the background (remember, they are just AI), have their eyes hung low, and their posture makes them seem to be subservient to the God-like figure they are facing. Overall, there seems to be little meaning to the image and can only be felt as some cruel depiction of socioeconomic status disparity –symbolized through clothing. It seems insensitive for Vogue to contrast golden and extravagant outfits with literal blue-collar uniforms; it is truly an image no one should have requested DALL-E to make.
Another explanation for this image might not be the fact that this was requested by the user but that DALL-E has an AI bias, as most people assume. DALL-E has already been criticized for racial and gender bias, especially with job professions. Lawyers, when requested, were only male, while flight attendants were always female.
Quite obviously, AI still has a long way to go. It seems like an innovation that could cut time for projects and make art creation more accessible, but it can just as easily steal people's art and promote biases in art. Therefore, we must responsibly approach AI to make sure that humans benefit more than they are harmed.
BY IZZI SCHULTE