“Many people enjoy AI art, but when human artists lose their motivation to create as machines take over their revenue streams, they run out of inspiration.”
Source: https://openai.com/dall-e-2/
As you scroll through social media, you are sure to find posts of countless images generated by artificial intelligence (AI). Tools such as DALL-E 2, Starry AI, and Jasper exploded in popularity, allowing users to create stylized versions of their own photos, leading to “John Lennon and red-headed cow sunglasses.” , users can receive results almost instantly without any further effort. Users flock to such websites because they are easy to use, free, and most of all, fun. Problem: Most of the time, AI “scrapes” the internet to get images, grabbing and using a large number of copyrighted images to figure out what certain words mean and what they are used in those images. Train yourself on which stylistic choices you have.
Naturally, the artist is upset. Andersen et al. v. Stability AI Ltd., etc. A possible class action lawsuit recently filed in the Northern District of California is due to AI art platforms being able to create images that convincingly mimic a particular artist’s style based on existing images of that particular artist’s work. , claims its AI platform is “siphoning commissions from artists.” themselves. “The complaint calls the same technology “just a complex collage tool.”
face the paradox
AI art raises some paradoxical themes in copyright law. First, what constitutes a “work” and what constitutes an “author”? Traditionally, a “work” has been defined as something that meets a minimum of originality and has been created by humans. Feist Publications Inc. v. Rural Telephone Services Co.499 US 340 (1991); Burrow-Giles Lithographic Co. v. Sarony, 111 US 53, 56 (1884); trademark cases, 100 US 82, 94 (1879). Last February, a U.S. Copyright Office review board confirmed that it had rejected his 2018 application to register works allegedly created using AI art.
But there is another doctrine in copyright law, which allows humans to be treated as “mere books”, tools by which “authors” achieve their artistic visions. SDNY has claimed that the filmmaker is the “author” of the footage shot at his direction by others. [] With a high degree of control over his meticulous artistic elements, so that the final product replicates his conception and vision, [work] should look like[.]” Lindsay v. Wrecked and Abandoned Ship RMS TITANIC, No. 97 CIV. 9248 (HB), 1999 WL 816163, *5 (SDNY October 13, 1999). This is because I was taking the image at his direction, no Considered the author of the footage they shot.
This creates a paradox. Humans with true intelligence are considered tools because they are smart enough to give instructions effectively, but AI (literally machines) are not treated as tools for use by their human creators. yeah. It’s not yet sophisticated enough to take complex directions. In other words, humans are smart enough to be treated like hammers.
As technology improves, legal hurdles rise
In particular, this comparison is only valid if the AI art generator requires trial and error to fix. AI technology has improved to the point where users can use sliders or similar value inputs to change certain elements of the resulting image, such as contrast, gamma correction, and brightness settings found on computer displays and televisions. You can change from the most basic options to the more creative ones. This analysis is debatable when esoteric options such as “scary”, “romanticism”, and “intensity” are involved, making it much more difficult to argue that there are enough human authors. At that point, AI Art will work like Protools, Photoshop, or Final Cut. In other words, it is a tool for humans to make creative choices regarding the final presentation of their work.
The difference between using the above program and using “AI 2.0” is input ownership. Infringing use of these programs is much clearer. If someone uses an unlicensed sample with Protools, tracking down the creator of that sample will be relatively difficult, since in almost all cases the individual sample was taken from his one source. it’s simple. When someone retouches a photo in Photoshop, the original photo can be traced. But AI art is conceptually different. It is possible to request a digital record of every image an AI “sees”, but the AI’s purpose is to synthesize vast amounts of information, not to retrieve and copy a single image. . In a sense, AI is working from artificial forms of inspiration rather than direct copying. for a single image.
This confuses analysis immeasurably. If a human were to represent in pop art a pointillist drawing of Washington crossing the Delaware after he spent 10 years in art school and visiting museums, that person is assumed to know the vocabulary, and to follow the rules. It is assumed that you know the rules well enough to break them. However, if an AI were to display millions of images online and create a “bowl of soup that is a portal to another dimension in the style of Basquiat,” it would be considered a possible copyright infringement. (All images are available on the DALL-E-2 homepage (https://openai.com/dall-e-2/) (last visited on January 29, 2023)).
All AI is fraught with moral and ethical conundrums, but under current copyright law, nothing protects an artist’s unique style, only an artist’s specific work. . Therefore, the potential for infringement lies in obtaining images, such as by downloading them, rather than using vast numbers of images to generate new works. If a human is put on trial for infringement, an effective defense would be, “But I didn’t copy one image of him, but some of his thousands.” .
Potential Paths for Protection
So where does that leave the artist? The copyright clause of the US Constitution permits the protection of intellectual property “to promote the progress of science and the useful arts.” US constant. art. I, § 8, cl. 8. Many people enjoy his AI art, but when human artists lose their creative drive as machines take over their revenue stream, their inspiration runs out. Therefore, if AI art generators are allowed to continue to exist, we need a system where artists are rewarded for their work. When the player piano was invented, Congress created a compulsory machine fee for composers who feared that sheet music sales would be diminished by the new technology. A similar approach may now be justified. The terms of user prompts and the number of artist images used in “training” AI will provide revenue to artists based on potential infringement based on making copies by downloading. This is already an established cause of action than facing the question of whether AI art constitutes derivative works. upon.
