/Imagine AI, Image Generation, Creativity
The phrase “/imagine” you see in the title may be a bit alien to you. This expression is actually used to command a tool called “Midjourney”, generating images via artificial intelligence (AI). The system creates pictures according to the expressions you add to the end by typing “/imagine”. In the screenshot below, you can see the image I chose from those produced by the command in the title.
Midjourney is just one of the AI tools made available to the masses. Today, AI is aiding people in many fields, from helping doctors with diagnosis to correcting the grammar of a business e-mail you write. Well, let me explain to you why I have started this article with Midjourney, whilst there are so many tools that can generate pictures from commands (semantic definitions).
With the rapid introduction of Web 3.0 and NFTs, discussions of digital property took on a whole new dimension. Pictures became the most visible face of this new “digital property”. Hence, rapidly popularizing image generation tools and many moral and economic debates that centered around them caught great attention. Most of the discussion is based on these systems’ inability to produce images “from scratch”. What they do instead is to generate new images by crossing those in their database with each other according to given commands. This brings up many problems surrounding originality and intellectual property.
In this article, I will try to evaluate the debates revolving around AI systems that can generate images through the concepts of originality and intellectual property and their relationship with creativity. To accomplish this, I will address the subject matter from the very beginning, the roots of humanity and its cultural history. I believe it also needs to be mentioned that volumes of corpus can be created on each of the topics discussed in this article. However, I will be brief in order not to lose focus.
The most intrinsic human ability: Imitation
Does language precede culture or is it the other way around? There is definitely something that precedes both: “plagiarizing”. As human beings, our strongest ability is to perceive something and imitate it. We created language through imitation of sounds in nature, pictures through the imitation of shapes and cutting tools through the imitation of claws…But are we the only imitators? Of course not! From monkeys to dogs and even crows, many creatures have the ability to imitate. So, what separates us from these animals?
We can mention two main distinguishing factors here. First, our ability to imitate is not pure repetition, but actually has a transformative aspect (creativity). Second, beyond simply teaching this imitation by “doing” it, we can preserve it by telling (language and communication). These two differentiating elements allow us to not only create culture but also make it progress. These two become the two main pillars that culture climbs on; innovation and cumulativeness. But how exactly?
Here are some ready-made texts
During the first years of my philosophy training, expressing my ideas was tough. Throughout the first years of my academic life, the phrase “X has an article on this subject, read it first” was enough to turn my dreams into nightmares. In addition to the excess of must-reads, the idea that nothing I wrote or thought could be original made me feel stressed and alienated me from my own work. If my ideas were already thought by someone before, how could I create value?
Over the years, this worry gave way to a realization. Philosophical thought was advancing cumulatively. Each philosopher developed new ideas by either opposing their predecessors or combining their ideas with others’. That’s why, in the works of Foucault, a 20th century philosopher, it was possible to find the ideas of Ancient Greek philosophers who lived more than 2000 years before him. In other words, Foucault’s thoughts were inspired by a history of thoughts dating more than 2000 years. This neither diminished Foucault’s contribution nor invalidated the Ancient Greek philosophers. The corpus was just building on itself.
The situation is not too different for the rest of humanity and its cultural history. Whether in the arts or sciences, cumulative progress is an everlasting phenomenon. Newtonian physics did not become invalid just because Einstein physics emerged. Handwriting did not disappear because the printing press was invented. People did not stop painting on canvas because it was possible to paint on the computer. Classical music did not become obsolete because electronic music was invented. On the contrary, just as Foucault owes his ideas to the history of philosophy, these advances that are emerging today owe their existence to their predecessors. Regardless of the value attributed to them in their relevant zeitgeist, all of the advances have been passed down from generation to generation to shed light on future steps in the memory of humanity and its cultural history.
So far, we have discussed cumulativeness and how it works. However, we still haven’t answered my primary concern. To what extent does this cumulative progress allow for originality?
Original is never finished
As humans, our ability to imitate is different from other living things, which is reflected in the multitude of expressions we use in its place: to be inspired, to mimic, to take example, etc. What all these concepts have in common is that they happen to be triggers of what we call creativity. So, it’s safe to say that originality is closely related to the way we imitate.
In short, no idea is isolated. For example, contemporary pop takes its roots from electronic music and hip hop, both having their roots in the genres that came before them, going all the way back to when we imitated rhythmic sounds in nature and created music. Likewise, while generating an idea, we build upon various bits of information and thoughts stored in our mind, in a personal manner. Our originality, then, comes from the way we bring these ideas together and from the effort we put into it. In short, how we “remix” is what makes us original.
Nothing is original. Steal from anywhere that resonates with inspiration or fuels your imagination. Devour old films, new films, music, books, paintings, photographs, poems, dreams, random conversations, architecture, bridges, street signs, trees, clouds, bodies of water, light and shadows. Select only things to steal from that speak directly to your soul. If you do this, your work (and theft) will be authentic. Authenticity is invaluable; originality is non-existent. And don’t bother concealing your thievery – celebrate it if you feel like it. In any case, always remember what Jean-Luc Godard said: ‘It’s not where you take things from — it’s where you take them to.
Jim Jarmusch
What makes the act of creativity original is that there is no limit to the number of ways we can bring things together. Airplanes can be produced via modeling a bird’s beak, a picture can be drawn with an inspiration from a stomach ache or a rock music concert can be given with a symphony orchestra. Likewise, any work can be reinterpreted, can be combined with one or more works, and can be brought together with elements from different cultures.
Let’s take as an example Hadise’s song “Superman” and its music video, which was produced in Turkey. When we look at the whole thing we can see that they managed to melt different global cultural elements in one pot. The song takes its name from the comic book hero that originated in the USA (obviously). The image of Superman being the strongest superhero is used correctly in the song in context of contemporary relationship culture in Turkey. The song, which can be categorized as pop music, has electronically sampled sounds of both western, modern and oriental instruments, and is digitally composed (probably) with additional electronic sounds. In the music video, there are inspirations from many other global cultural elements. Hadise wears an outfit similar to the cape that Kylie Minogue wore in the “In your eyes” music video. Her dance moves are very close to Beyonce’s style. In one scene, there is a move that Hadise makes on a chair, which seems to be a reference to the one Demi Moore made in the movie “Striptease”.
For translated song lyrics: https://www.musixmatch.com/lyrics/Hadise/Superman/translation/english
Obviously, I didn’t choose this example by chance. At the time of the song’s release, there was a very intense discussion of plagiarism around Turkish songs and especially music videos. Even if I don’t remember whether this song had its share from those discussions, it shouldn’t be hard to guess why it might. So, at what point does creativity really turn into stealing?
No trying the already tried
In order to understand whether an act of creativity involves theft, we must first look at what is being stolen. This is the point where the concept of “intellectual property” emerges. Intellectual property is something that you create using your mind, like an artwork, story, or invention. Legally, just like physical property, it can be sold, transferred or owned by more than one person. But hold on, haven’t we just said that no idea or invention is independent of humanity and its cultural history? How is it then possible to claim ownership of any mental production?
In fact, we have given the answer to this question earlier while defining originality. Surely, no intellectual production takes place in isolation. However, what makes an idea special to us is the way we put it together. Well, if we assume that we will produce in our own way while doing intellectual production, how can creativity morph into stealing?
At this very point, leaving the definitions of laws aside, I will conclude the reasoning we have been building throughout the article. If a work is reproduced directly, without the consent of its producer, without the act of “remixing” we’ve mentioned above, and with means of turning this reproduction into a personal profitable commodity, creativity has turned into theft.To put it simply, if you copy a work or idea as is, without adding any interpretation, and aim to gain personal benefit -which won’t be transmitted to the owner of the intellectual property in any form- you are stealing. For example, printing and selling an artist’s work on a t-shirt or using a part of a doctoral thesis directly from someone else’s work is stealing.
Of course it is possible to talk about a few controversial practices brought by the modern world. A good example of this can be the movements on social media called “challenges” which are based on the reproduction of the first idea created by other users. Likewise, the practice of cutting pieces from the works of other artists and adding them to your own songs, which we call “sampling”, is also very popular. The music industry is way ahead in terms of intellectual property protection, as they have been pondering about it for a long time. Therefore, they have implemented systems that enable the owners of copyrights to claim their shares in such use cases. Likewise, the world of painting, which had the practice of “reproduction” since forever, has started to get its fair share from these technological developments with AI image generation tools. So, in the light of all that we’ve been talking about, how can we evaluate the possibilities offered by this new tool in terms of creativity and intellectual property?
/Imagine paints, not a painter, interprets someone else’s idea
The image above is from Midjourney’s Discord server. I don’t know the prompt that created the picture. However, guessing from the message and the picture itself, it was created randomly by combining the styles and works of two artists, Lisa Frank and Henri Matisse, which I think was a success. On the face of it, it seems to meet many of the needs we can expect from a painting that can be described as a work of art. It has a certain style, a composition, and it conveys an aesthetic feeling.
Below is Moebius’ reinterpretation of Edward Hopper’s famous painting “Nighthawks.” It meets all the aforementioned criteria, including combining another artist’s work and style with someone else’s. So what separates these two from each other?
At this point, looking at how AI creates these pictures can guide us. As far as we know, these systems train themselves, just like human painters, by studying the work of other artists and trying to draw similar things. They do this by going through an archive of more than 3 billion images. Moreover, they do this million times faster and with more repetitions than a human painter. Pictures, all collected from the internet, are not archived with just visual data, but each image is accompanied by explanations added by the person who uploaded it (You can access a part of the database that includes 12 million images used by Stable Diffusion from this link, and the details of the database from here). Whenever you give a prompt, through its algorithm the artificial intelligence creates random images. It does this by combining the images that have descriptions with the words in that prompt, alongside with feedback from its previous creations. As people choose the best amongst these pictures, the algorithm continues to improve itself based on people’s preferences.
At this point, the main distinction -in a narrow scope- can be made through how this system works. If we were to remove its database, AI wouldn’t be able to create. On the other hand people began painting by interpreting nature, when there were no pictures around. I’m not going to go into a lengthy discussion about how humans are actually like machines and technically do what AI is doing via its database through their memories. Instead, I will try to analyze what sort of rights those who use this tool have over what they produce.
This is not a painting
The Treachery of Images by Rene Magritte
Let’s imagine a scenario: You share with a painter friend, whose talent and experience you trust, an idea that you think she can turn into a painting, and she goes and does that. Who made this painting? There is no doubt that you contributed to the creation of the work, but can you claim that it’s yours? Unless the painting is a conceptual artwork like René Magritte’s “Treachery of Images”, I don’t think so.
If we were to keep contemporary conceptual art in a separate place, the main elements of a painting consist of mostly visual elements, such as how the objects in the painting are drawn and how colors are used together, rather than what sort of stuff comes together in the painting. We can think of these essential elements as craft and visual expression abilities.
For example, Édouard Manet’s “Luncheon on the Grass” would be a work of no importance to anyone if it were not for Manet’s ability and experience to put this composition together, even though it contains Manet’s critique of Renaissance Art. Therefore, what makes this painting a work of art is not only the renaissance criticism, but the fact that it has created an original and aesthetic value through craftsmanship.
Taking all this into consideration, even if we argue that pictures produced by AI are works of art, we can’t say that the person who created that painting, that is, the work of art, is the person who commands the artificial intelligence. Because, although the person who creates the prompt decides the elements that bring the composition together, and perhaps whose styles are to be brought together, it is the AI that transforms it into a painting. Therefore, it doesn’t seem possible for users to claim intellectual property on these works, at least as works of artistic paintings.
The idea that intellectual property belongs to the tool that generates this composition is also questionable. Although tools like Midjourney transfer the intellectual property rights to users for a certain fee, there are a few problematic points here. First of all, the capabilities of these systems come not from themselves, but from the paintings of the artists in their database and from the people who give feedback to the system. Because of this, prompts such as “in the style of X” and “in the style of Y” usually create the most successful results. Therefore, there is an unauthorized use of this talent together with the work itself, rather than the inspiration of a talented person interpreting the work. As a result, the thing that separates the picture, which brings together the styles of Lisa Frank and Matisse, from Moebius’s interpretation of Nighthawk is that Moebius’ talent is unique, whereas Midjourney’s talent belongs to Lisa Frank, Matisse and the whole users who trained it. Thereby, systems that produce images through AI do not seem to be able to provide the authenticity approach we have been defining that makes intellectual property possible. This makes it controversial for images produced by this tool to achieve the status of artistic paintings that can have an intellectual property.
To conclude, this issue, starting to attract much more attention in the context of the digital property problem brought upon by Web 3.0 and NFTs, has moral and economic implications. At this point, it is obvious that the painting industry is going to have to follow the footsteps of the music industry, which has been dealing with these issues for a long time. The question is how these artists, many of whom are independent and do not have big profit-oriented labels behind them, can walk this path to protect their authenticity. One thing is certain, these technological advances will cause radical changes in the lives of artists. The real question to be asked is, will these changes cause the art of painting to steer into a different direction like it did after the invention of the camera, or will it force it to adapt to this change?