‘The quest for artificial intelligence is a quest for intelligence itself.’

Spatial AI Ecologies explores creative AI in a collaborative fashion to produce a new cognition of interiority. Intelligence, in its various forms, is an emergent property of material assemblages that are ontologically diverse, just like their functions. Following that, we will be categorizing intelligences here as biological or mineral - as opposed to natural and artificial.

Tracing down the line, it is geology that gives rise to biology, and it is biology that produces technology - using geology, minerals. Through this process, intelligence is plural and is achieved through complex electrical amalgams of matter, organic and non-organic.

After that point, things stopped being linear, where technology and biology simultaneously consumed all of the above to produce new geologies, biologies, and technologies. Spatial intelligence is at stake in this project, meaning creating an artificial condition of interiority using collaborative forms and workflows of AI models.

From caves to virtual spaces, interiority is vital, and the aesthetic is disturbingly analogous, as our senses haven’t changed much. Paintings on cave walls, fellow inhabitants, fauna, shadows projected by that project shadows, soundscapes, texture, material, and more have framed what interiority is. Therefore, we explore a workflow deeply based on ai-algorithms to generate space, its textures, and interior elements that come together to form a spatial microecology of human and machine-made elements.


Project AI Description

The project is the yield of several AI algorithms to create the diverse elements of the ecology. We produced meshes, images, and audio, as a continuum that forms the artificial spatial condition. The AI hence was used as a new agent for creating interiorities and as an aesthetic in the spatial ecology rendered.

Text-to-Image algorithms were used primarily to prepare data for the Image Mesh algorithms, reducing the human factor within the workflow. Image to Mesh was used to produce the main space, or enclosure, the population, digital flora, and other objects. Digital human avatars - meshes- were generated using a different image-to-mesh model listed below. In parallel, VQ GAN + Clip was used to generate a sequence of images that were made a video and mapped as a digital material of the mesh. Finally, music was generated using VQ-VAE as a spatial soundscape.