Author: Joshua Saji
Topic: AI Art Debate
Within the past five years, Artificial intelligence has taken the world by storm. Today, it’s an integral part of many people’s daily lives, as people take advantage of the various chatbots to perform menial, everyday tasks. However, arguably the most interesting aspect of these models is their capability to generate art. With just a few words, a user can have an AI chatbot produce an aesthetically pleasing piece of art based on their specifications. While this certainly seems like a remarkable achievement in technology, there are numerous ethical considerations surrounding the discussion of generative art. Who owns the art piece? What are the copyright considerations involved? However, the topic we’ll explore today is whether AI-generative art can truly be considered art.
Origins of Generative Art
Before we delve into this question, we need to discuss the origins of generative art, starting with Harold Cohen in the 1960s. Cohen was a computer scientist and artist, and was a professor in the Visual Arts department at the University of California – San Diego. Most notably, however, he was known for his creation of AARON, recognized today among computer scientists as the first artificial intelligence model to achieve generative art.
The way AARON works makes it stand out strongly from other models capable of generative art. In particular, AARON isn’t a piece of software – it’s a small device that acts like a human drawer. Cohen built his own plotters that could translate machine instructions into lines drawn on paper. He also included brushes for AARON to add color to its generated art. On the other hand, models such as Midjourney or DALL-E are strictly software, where any user can prompt the model with their desired illustrative piece, and the model will produce it based on the user’s needs. Through the prompt, the model takes millions of images of illustrations as data points, feeds them into its mathematical model, and then spits out its own piece.
Discussions about generative art will naturally delve into the true nature of it. It’s safe to say that Harold Cohen’s beliefs on what constitutes art heavily influenced his creation of AARON. In his essay, “Parallel to Perception”, he believed that among everything that humans have made, the computer had the most potential to perform tasks parallel to the capacity of the human mind. One of these tasks included generating art autonomously. At the same time, he believed that art’s significance was beyond the superficial, visual aesthetics, and laid within the meaning the artist put forth in the creation of their pieces. These views can be reflected in AARON – although the artmaking process is largely removed from the artist, it is still up to the artist to have a vision and intended meaning behind the piece they want. Whereas in models such as DALL-E, meaning or any sort of vision behind the piece isn’t necessarily included or required.
However, it’s not just meaning that sets human-generated art apart from AI-generated art. In fact, one may still argue that meaning may not be fully lost with AI, so long as the user of the generative AI model purposely includes the meaning within their prompts. But, arguably the most important aspect of human-produced art, and creativity in general, is the fact that ultimately, any intentional decisions behind the art they produce, or any inspiration they find, or any meaning or vision they want to convey through art, stems from nature and direct observations of the physical world.
This crucial aspect of art is what artificial intelligence models lack. The biggest observations that these models make are from the millions of previously-produced works from artists throughout human history. Outside of that, these models solely rely on mathematical instructions and data provided by their developers. They have no true senses or experiences of the same world that artists live in. It’s equivalent to someone claiming to have visited Paris because they’ve seen pictures online – they may have an idea of what the city looks like, but they’ve never experienced it for themselves and never will until they actually step foot into it.
To explore this idea even further, we can take a look at ChatGPT’s new feature which allows users to produce graphics in the style of films made by Studio Ghibli. This new capability has taken social media by storm, with influencers, celebrities, and active users posting portraits, iconic movie scenes, memes, and more – all under this style.
ChatGPT & Studio Ghibli
Studio Ghibli is an animation studio based in Tokyo, Japan that creates films. The studio is directed by Hayao Miyazaki, and is most noted for its heavy use of traditional drawing techniques. Each frame of every film the studio has produced is hand-drawn by Miyazaki’s team of artists, leading to the colorful, smooth and vibrant films that millions of viewers enjoy today. However, the beauty behind Ghibli films, and much like millions of art pieces created in human history, doesn’t lie solely in its visuals – rather, it’s emotional and captivating storytelling. The visuals only help make the director’s vision come to life.
There’s no doubt that ChatGPT is capable of producing stunning visuals. DALL-E, the AI artist that’s integrated into ChatGPT, has been shown to create pieces possibly beyond what the human mind could create. ChatGPT’s Ghibli-style production is certainly no exception. Yet, no matter how stunning the visuals may appear, ChatGPT’s generative art capabilities should not be treated the same as the human’s capabilities.
Firstly, ChatGPT does not, in any way, shape or form, follow the principles laid out by Miyazaki for his films, where each frame is hand drawn. Much like how other AI models generate art, ChatGPT simply takes a set of data points, and outputs its own piece based on the user provided prompt.
Secondly, and as discussed above, the pieces the model produces don’t contain intrinsic meaning. As discussed above, and noted by Harold Cohen, the meaning behind something visual is critical. Studio Ghibli’s films used art as a conduit to convey the vision outlined by Miyazaki and his teams. Every decision behind the creation of Studio Ghibli’s films were done to convey some sort of meaning to the viewers. There isn’t an active decision-making and creative process behind ChatGPT when creating these pieces. The art produced by ChatGPT lacks the charm that the Studio Ghibli films have.
ChatGPT’s intervention with Studio Ghibli and Miyazaki is far from anything we haven’t seen before. In fact, Miyazaki strongly condemned the use of AI almost a decade ago. A team of developers developed a machine learning model to create unique, animated character movements that may be difficult or even impossible for humans to replicate. Some of these movements were grotesque and were expressive of agonizing pain, which could apply not only to Studio Ghibli films, but other modes of media, such as video games. When this team met with Miyazaki, he took grave offense, as he believed that AI will never capture the essence of human pain.
In Summary
The issue of AI-generated art doesn’t only lie in visual pieces, such as paintings, drawings, or animated films. One could extend this question to other forms of art or modes of human expression and creation, such as literature and music. Regardless, the point still remains – whether AI can generate novels or exquisite illustrations, they still lack the soul and humanity that human-curated art has. There’s no intrinsic meaning behind what a computer generates or what it could generate, and there never will be. And if we continue to treat computer and human-generated art as the same, then society will surely lose its own humanity.
References
“Harold Cohen: Aaron.” Whitney Museum of American Art, https://whitney.org/exhibitions/harold-cohen-aaron.
Pogrebin, Robin. “Harold Cohen, Pioneer in AI-Generated Art, Is Having a Renaissance.” Art Basel, https://www.artbasel.com/news/harold-cohen-ai-programmer-artist-technology?lang=en.
Cohen, Harold. Parallel to Perception. https://www.aaronshome.com/aaron/publications/paralleltoperception.pdf.
Dwoskin, Elizabeth. “ChatGPT’s Ghibli-Style Images Raise Questions of Copyright and Creativity.” The Washington Post, 28 Mar. 2025, https://www.washingtonpost.com/technology/2025/03/28/chatgpt-ghibli-ai-images-copyright/.
“AI Animation vs. Miyazaki.” YouTube, uploaded by Great Big Story, https://www.youtube.com/watch?v=ngZ0K3lWKRc.
Andrews, Mike. “The Painting Fool and Aaron: Creative AI in the Art World.” New Atlas, 20 Feb. 2015, https://newatlas.com/creative-ai-algorithmic-art-painting-fool-aaron/36106/.