Towards the end of writing this, OpenAI released a new image-generating model. My Social Media was flooded with images generated in the animation style of Ghibli Studio. Uploading a photo, a person, a piece of memory, or a picture of anything into ChatGPT, accompanied by the prompt "Redraw this photo into Studio Ghibli style animation." Voilà, a style once revered as a pinnacle of hand-crafted animation, was now accessible at the click of a button. Hayao Miyazaki, the Co-Founder of Ghibli Studio, once described AI-generated images as "an insult to life itself" in a documentary while researchers showed him some footage of AI-generated animations.
This OpenAI update sparked a significant debate, especially in the East Asian art scene. Many creative workers not only condemned the developers behind such technologies but also criticized the users, whom they saw as complicit in the theft of artists' hard-earned labor and stylistic signatures. For them, the mass reproduction of distinctive visual styles felt like a violation of cultural and artistic authorship. On the other hand, public users responded with enthusiasm, arguing that these new tools democratize creativity. To them, the ability to generate images in revered styles represented another step in the decentralization and devolution of technique, where anyone could participate in artistic expression.
I found myself caught in the midst of these debates. On the one hand, I hesitated to criticize those who projected their emotions and memories onto AI-generated recreations. Who am I to dismiss an attempt to reclaim or reinterpret their past through a medium newly made available? On the other hand, the making of Ghibli Aesthetic, like countless others, is by no means an intentional design; it rather emerged from neural networks, scraped, inferred, and statistically embedded in massive datasets. Machines never had the intention to replace human artistic expression; it is just there, existing, while we impose our intention on it.
Still, these processes still involve human decisions at every level. From the scientists who designed the architecture to the engineers who made the program to label training datasets to the users crafting prompts, it is still people who shape the outcome. The Ghibli style did not emerge from nowhere; the so-called creativity of the model was built atop a collective archive of human labor, intensive calculations, and cultural production. While it may appear that the machine "creates," it is still important to remember that it does so through a diffused and collective authorship mediated by companies, platforms, and infrastructures that prioritize scale and automation.
The question, then, is not simply whether this is art, or who is responsible for it. It is about how we choose to engage with the collective systems we are all, in some degree, complicit in. In the context of artistic production, the challenge is then clear: how do we acknowledge and navigate this entanglement, between individual intention and systemic design, between authorship and infrastructure?
Since semiconductor productions went to headline, like every other artists I look for references, and realized that artists have increasingly turned their attention to the infrastructures that power such technologies; The global semiconductor industry has become a rich topic of inquiry. Sam Ghantous's series How to Turn the Earth Inside Out uses UV prints on silicon wafers to depict modernist architectural forms, collapsing the image and medium into a reflection on the material underpinnings of computational vision1. In his recent exhibition Your Golf Course Made My GPU, he expands this critique, linking the aesthetics of microchip design with global land use and geopolitical tension. Hong Kong artist Nadim Abbas, in his large-scale installation Pilgrim in the Microworld at the 2023 Taipei Biennial, constructed a pseudo-integrated circuit out of sandboxes to discuss technological complexity and military architecture2. Taiwanese artist Su Yu-Hsin made Particular Waters, a video installation tracing the water infrastructure in Hsinchu, the city where TSMC is located3. I see this not just as a trend, but as a necessary step in artistic inquiry, where the aesthetics of image-making are inseparable from the material conditions that make them possible.
As I reflect on my own practice and on the writing. I no longer see engagement with generative AI as a question of approval or refusal. Rather, it is about finding a position somewhere between fascination and critique, between participation and resistance. To work with this tool is to enter a network of culture, computation, and infrastructures, where every output carries not only its aesthetic significance but also ecological, political, and historical weight. The machine does not dream; it does not even create with intent. And yet, in its outputs, we project memory, desire, and intent. AI is thus a mirror, a reflection of the data we train, the systems we sustain, and the infrastructures we overlook.
To work with AI, as an artist, is to work with these contradictions, to remain close not only to the surface of the image, but also the condition that shaped it, to the layers beneath, the labor behind, the heat in circuits, and to the silence hums behind the interface.