In recent years, artificial intelligence (AI) has advanced rapidly. This has led to the creation of what is now known as AI-generated art, which has sparked conflict within art communities over what exactly AI art could mean for human artists. But what is AI art? And how does its existence affect not only the way our society views art but also the way we define art as a whole?
Artificial intelligence text-to-image programs such as DALL-E, Adobe Firefly, and Midjourney are a form of generative AI systems called diffusion models. They use patterns found in big sets of data to create images based on text prompts. In other words, these systems will take information, find associations between text prompts and images, and use those patterns to make an image that will fulfill a prompt. For example, if a generative AI system found in a data set that the word tree was associated with an image of a tree, it would then use that pattern to create images of trees when prompted.
The nature of AI generators is what creates the greatest debate when it comes to AI-generated art - the legality of training artificial intelligence on copyrighted material. Copyright law, when it comes to visual art, is already very convoluted. The most important thing to note is that when an artwork is created in a fixed form, it is protected under copyright law. Registering an artwork with the U.S. Copyright Office is not necessarily required, but it allows for a stronger argument in copyright infringement cases. However, with AI art comes a few complications.
First of all, many believe that AI art cannot truly be copyrighted as it isn’t created by a human artist. Several cases have been filed around the U.S. where people have tried and failed to copyright AI artworks. An example of this was the case filed by the CEO of Imagination Engines, Stephen Thaler, who was denied a copyright application for his AI-generated art. The case was decided based on a ruling from 1884 when it was argued that photographs could be copyrighted since cameras were only considered a tool to create art with, not the ultimate creator. However, this decision is likely to not be the end-all-be-all decision over the copyright of AI-generated art. Thaler argued that, in this case, the copyright law could be loosely interpreted, and that the most recent version of U.S. Copyright Law could extend to cover AI-generated art as well.
Another issue with trying to copyright AI-generated art is the possible copyright infringement that is being committed using these AI generators. As stated before, these systems only work because of the data they are trained on. For example, in the case Stability v. Andersen, a couple of artists filed a copyright lawsuit against Stability AI, Midjourney, and other companies that allegedly used copyrighted images to train their AI. Originally, Illustrators Sarah Andersen, Kelly McKernan, and Karla Ortiz filed a case against these companies, but the judge, U.S. District Judge William Orrick, found flaws in their argument. Orrick argued that the AI output did not directly infringe copyright, but he kept the idea that the AI training did violate their rights. Ultimately, whether or not AI-generated art can truly be copyrighted or be considered copyright infringement is yet to be decided. Copyright laws have always been and will continue to be based on interpretation, and the decisions in these cases may change over time as society’s view on copyright and AI-generated art changes.
The stir that AI-generated art has created has not only impacted the legality surrounding the creation of art but also the people who create art. Depending on the artist you ask, opinions on generative AI can range from it being an innovative new tool for creating artwork to it being the death of creativity. In an article written by the Harvard Gazette, several artists were interviewed for their perspectives on AI art. While they all had varying opinions, the general belief they all held was that AI art could not, in its current form, be a complete replacement for human art. Yosvany Terry, a composer, lecturer on music, and director pointed out that, unlike humans, AI cannot express emotion or work improvisationally. Art like jazz is something that is created based on a person’s reactions and emotions, which at this current moment is not something AI can replicate. Architect and urban planner Moshe Safdie also pointed out that while AI can be used to help create the technical aspects of art, it cannot mimic the eye artists have for aesthetics or emotion. To put it bluntly: AI generators, in their current capacity, cannot mimic the way artists put life into their art. They don’t have the same curiosity and creativity that humans do, so they cannot replace human art. However, this doesn’t mean that they are entirely useless, as many of the artists pointed out that AI generators could be used to enhance the abilities of artists.
When all is said and all is done, the greatest and arguably most important impact that AI generative art has had on our society is altering how we define art. The definition of art isn’t by any means steadfast – in fact, its definition could change depending on who you ask. To some, art is defined by the emotion communicated through it. To others, art is simply a painting on a wall or a drawing in a sketchbook. Perhaps that’s the main reason why AI-generated art puzzles our society, and why it has impacted us so deeply. AI-generated art could be art, but it also could simply be a tool, similar to a camera or a pen. It all depends on the definition you give art. That is why it’s so hard to categorize it in our current copyright systems for artwork and why it’s distressing to some artists, and life-changing to others.
AI-generated art isn’t going away anytime soon, and it will only continue to become more advanced and developed. One day, AI art may be able to fully replicate the emotion and creativity that humans have. Its existence will likely continue to impact our society. Whether or not that impact is positive or negative will depend on how we use it going forward and how we perceive it.
Title Image: “Midjourney: 25 variations of an industrial warehouse” by kevin dooley
Adobe. “What is AI art & how does it work?” Adobe, September 26, 2023. Source
Brittain, Blake, and Barnes Cellino. “Artists take new shot at Stability, Midjourney in updated copyright lawsuit.” Reuters, November 30, 2023. Source
Mineo, Liz. “Is art generated by artificial intelligence real art?” Harvard Gazette, August 15, 2023. Source
Naughton, John. “Can AI-generated art be copyrighted? A US judge says not, but it’s just a matter of time.” The Guardian, August 26, 2023. Source
U.S. Copyright Office. “Visual Art.” U.S. Copyright Office, August 2019. Source
An Inside View of Girls with Leadership Roles