Artificial Intelligence (A.I.) has revolutionized many industries and has made significant advancements in recent years. But as technology advances, whether A.I. is stealing from artists has become increasingly relevant. With the ability to create realistic images, music, and even written works, A.I. has the potential to replace human artists. But is this a fair representation of the creative process?
Last year, Kelly McKernan, an artist based in Tennessee, discovered that their name was being used more and more frequently in the realm of A.I.-generated images. McKernan is known for their paintings of female figures with a dreamy look, often rendered in an acid-hued style mixing Art Nouveau and futuristic elements.
The August edition of Metaverse Post published a list of hundreds of artists, which, when fed into an A.I. generator, could be used to create art inspired by “Lord of the Rings.” Each artist was chosen based on their work’s evocation of genres such as anime, modernism, and “Star Wars.” Kelly McKernan was one particular artist who made the list.
McKernan noticed that participants on the Discord chat for Midjourney, an A.I. generator, had mentioned their name more than twelve thousand times in open invitations.
The resulting visuals – owls, cyborgs, gothic funerals, and alien motorcycles – were strongly reminiscent of McKernan’s art. They informed me that things had gotten “weird” at this point, and it was beginning to look pretty close to their work as if it had been studied and combined with other people’s works. They could recognize their style in these images.
McKernan, Sarah Andersen, and Karla Ortiz are plaintiffs in a class-action lawsuit brought by attorneys Matthew Butterick and Joseph Saveri against Midjourney and two other A.I. imagery generators – Stable Diffusion and DreamUp – that was filed last month.
The three models all leverage laion-5B, a publicly accessible non-profit repository that contains over five billion pictures from the web, including artwork by several creators. Other tools, such as dall-e, are also based on the same principles.
Butterick informed me of the supposed misconduct, which he referred to as “the three ‘C’s”: The artists had not approved their copyrighted artwork to be included in the lion database; they were not reimbursed for their involvement, despite companies like Midjourney demanding fees for utilizing their tools; and there was no recognition of their influence when A.I. images were generated using their work.
According to Butterick, when creating a picture, these generators will “give it to you as if it is free of copyright”; however, he explained that each image produced by the generative tool would be an illegal and derivative work.
It can be tricky to make copyright claims based on style. In visual art, courts have been known to side with the copier rather than the copied. An example of this is a court case in 2013 involving artist Richard Prince and photographer Patrick Cariou. Prince had taken photographs from Cariou and incorporated them into his work; however, due to transformative use, it was decided that no infringement had occurred as Prince had substantially changed the source material.
Recently, the music industry has been more inclined to be strict in its decisions. In 2013, Robin Thicke and Pharrell Williams were taken to court by the Marvin Gaye estate for allegedly copying parts of his song “Got to Give It Up” in their work entitled “Blurred Lines.” Ultimately, they lost this case.
In her recent essay on Butterick and Saveri’s suit published on her website, intellectual-property lawyer Kate Downing suggested that the A.I. image generators might be more similar to the former than the latter. According to Downing, “It may well be argued that the ‘use’ of any one image from the training data is…not substantial enough to call the output a derivative work of any one image.” She explained, “Mathematically speaking, the work comes from everything.”
Butterick and Saveri claim that A.I. generators do not provide transformative use but rather just a mechanized “blending” of the source material. As Butterick put it, they are not attempting to prosecute image by image but instead challenging the entire technique behind the system.
Last week, Getty Images filed legal action against Stable Diffusion, claiming that their utilization of Getty’s stock photography is a “brazen infringement” to an overwhelming degree. Despite any potential legal merits, such allegations have a certain moral weight. It must be remembered that Artificial Intelligence generators can only function due to the hard work of people like McKernan, who unintentionally supply content. Therefore, the litigators are not alone in this matter.
In his 2013 book “Who Owns the Future?” Jaron Lanier, a technology critic and philosopher, expresses that digital information is essentially people in disguise. A spokesperson from Stability AI, which created Stable Diffusion, responded to a lawsuit by stating that it was due to a lack of understanding of generative A.I. technology and copyright law; however, no additional explanations were given.
As we’ve discussed, the debate around the impact of Artificial Intelligence (A.I.) on the art world is a complex and multifaceted one. On the one hand, A.I. has the potential to create incredibly realistic and impressive works of art. Still, on the other hand, it raises questions about human creativity’s role and originality’s value in the arts.
Source: newyorker
