Artificial Us is an immersive, interactive installation that interrogates the cycles of extraction, transformation, and return that shape both our landscapes and our technologies. Built in TouchDesigner and powered by Kinect body tracking, the piece invites visitors to navigate a series of evolving 3D environments—each generated in real time from AI-created images and depth maps.


Conceptually, the work examines how neo-extractivism operates across both physical and digital realms. It traces the path of resources like lithium from the salt flats of northern Chile and Argentina, through sites of extraction and labor, across oceans, and into distant centers of industry. This journey is not only about materials, but also about the stories and identities that are lost or distorted as they move further from their source—a process mirrored in the way AI models are trained on data from the Global South, yet often fail to represent these places accurately.


The creation process begins with personal photographs from Chile and Argentina, which are used to train custom LoRA models on Stable Diffusion. These models generate new images that reflect the textures and realities of these landscapes and communities. The original photographs and the AI-generated images are presented side by side, inviting viewers to consider what is preserved and what is transformed in the act of generation.



To build the interactive 3D environments, each AI-generated image is processed with Depth Anything V2 to extract a depth map. In TouchDesigner, these depth maps and images are combined to create point clouds, forming immersive spaces that can be explored in real time. The Kinect sensor tracks the audience’s movements, while custom Python scripts control scene transitions and user interactions, ensuring a seamless and responsive experience.


As the narrative unfolds, the visuals transition from highly specific, fine-tuned scenes to more generic, distorted environments generated by base models. This visual shift underscores the loss of specificity and meaning that occurs through cycles of extraction and abstraction. Accompanied by a written and spoken narration, Artificial Us critiques the power structures embedded in both resource extraction and AI technologies, while also exploring how these tools might be reclaimed to tell more representative and meaningful stories.



Artificial Us is being developed in collaboration with Tiago Aragona (Argentina), as part of our MFA thesis at Parsons School of Design.