Step inside the mind of a generative AI, a vast mathematical space where all the memories and stories we shared online are compressed into a statistical resource. This data-driven VR experience reveals how AI models extract and organize our visual culture into a multidimensional geography of meaning and how the space between datapoints promises infinite potential. But this digital mirror isn't neutral, it's curated, filtered and biased toward aesthetic ideals. As you navigate through dense clusters of contemporary online culture, it poses a critical question: can this really be a space for artistic discovery or are we pulled towards the points of maximum statistical gravity?
Terra Ultima is a data driven experience that explores the concept (and promise) of Latent Space as a contemporary cultural phenomenon. Latent Space is a metaphor used in data science to describe the empty space between data points. It is what allows generative AI to extract “new” images and texts seemingly out of nowhere. Because of the scale and abstraction of AI technology it is often considered a “black box”. Something comes out, but we don’t know how. Terra Ultima uses immersion to descend into this hidden world, translating the abstract mathematical space of an AI model into a virtual environment. It shows the training data that was appropriated by the companies who built and trained these models (through a dataset called LAION) and allows the viewer to explore this space in an embodied way. Along the way it raises the question; can this really be a space for (artistic) discovery or are we destined towards endless variations of the same thing?
Interaction
The experience is based on a data visualization of the latent space of Stable Diffusion (a popular open source generative AI). A narrator introduces the space and reflects on the cultural and political dimensions of generative AI. The viewer can point their controller to explore the images, how they relate to each other and where they came from. By pulling the trigger button, the viewer can move from one cluster to the next. Since the experience is data driven and retrieves the images from the internet in real time, a stable internet connection is required. Only one controller is needed. Both controllers may be held but only one is active at a time (Supports both right- and left-handed interaction).
Run time 8 minutes.
Terra Ultima is a WebXR experience designed for headset VR.
Step inside the mind of a generative AI, a vast mathematical space where all the memories and stories we shared online are compressed into a statistical resource. This data-driven VR experience reveals how AI models extract and organize our visual culture into a multidimensional geography of meaning and how the space between datapoints promises infinite potential. But this digital mirror isn't neutral, it's curated, filtered and biased toward aesthetic ideals. As you navigate through dense clusters of contemporary online culture, it poses a critical question: can this really be a space for artistic discovery or are we pulled towards the points of maximum statistical gravity?
Terra Ultima is a data driven experience that explores the concept (and promise) of Latent Space as a contemporary cultural phenomenon. Latent Space is a metaphor used in data science to describe the empty space between data points. It is what allows generative AI to extract “new” images and texts seemingly out of nowhere. Because of the scale and abstraction of AI technology it is often considered a “black box”. Something comes out, but we don’t know how. Terra Ultima uses immersion to descend into this hidden world, translating the abstract mathematical space of an AI model into a virtual environment. It shows the training data that was appropriated by the companies who built and trained these models (through a dataset called LAION) and allows the viewer to explore this space in an embodied way. Along the way it raises the question; can this really be a space for (artistic) discovery or are we destined towards endless variations of the same thing?
Interaction
The experience is based on a data visualization of the latent space of Stable Diffusion (a popular open source generative AI). A narrator introduces the space and reflects on the cultural and political dimensions of generative AI. The viewer can point their controller to explore the images, how they relate to each other and where they came from. By pulling the trigger button, the viewer can move from one cluster to the next. Since the experience is data driven and retrieves the images from the internet in real time, a stable internet connection is required. Only one controller is needed. Both controllers may be held but only one is active at a time (Supports both right- and left-handed interaction).
Run time 8 minutes.
Terra Ultima is a WebXR experience designed for headset VR.
Step inside the mind of a generative AI, a vast mathematical space where all the memories and stories we shared online are compressed into a statistical resource. This data-driven VR experience reveals how AI models extract and organize our visual culture into a multidimensional geography of meaning and how the space between datapoints promises infinite potential. But this digital mirror isn't neutral, it's curated, filtered and biased toward aesthetic ideals. As you navigate through dense clusters of contemporary online culture, it poses a critical question: can this really be a space for artistic discovery or are we pulled towards the points of maximum statistical gravity?
Terra Ultima is a data driven experience that explores the concept (and promise) of Latent Space as a contemporary cultural phenomenon. Latent Space is a metaphor used in data science to describe the empty space between data points. It is what allows generative AI to extract “new” images and texts seemingly out of nowhere. Because of the scale and abstraction of AI technology it is often considered a “black box”. Something comes out, but we don’t know how. Terra Ultima uses immersion to descend into this hidden world, translating the abstract mathematical space of an AI model into a virtual environment. It shows the training data that was appropriated by the companies who built and trained these models (through a dataset called LAION) and allows the viewer to explore this space in an embodied way. Along the way it raises the question; can this really be a space for (artistic) discovery or are we destined towards endless variations of the same thing?
Interaction
The experience is based on a data visualization of the latent space of Stable Diffusion (a popular open source generative AI). A narrator introduces the space and reflects on the cultural and political dimensions of generative AI. The viewer can point their controller to explore the images, how they relate to each other and where they came from. By pulling the trigger button, the viewer can move from one cluster to the next. Since the experience is data driven and retrieves the images from the internet in real time, a stable internet connection is required. Only one controller is needed. Both controllers may be held but only one is active at a time (Supports both right- and left-handed interaction).
Run time 8 minutes.
Terra Ultima is a WebXR experience designed for headset VR.